Fire-fighting pipe network monitoring data denoising method and system

Abstract

The invention discloses a firefighting pipe network monitoring data denoising method and system, and the method comprises the steps: carrying out the variational mode decomposition of firefighting pipe network monitoring data, and obtaining a plurality of intrinsic mode functions; for each intrinsic mode function, optimizing a parameter of a non-local mean algorithm to obtain an optimal denoising parameter corresponding to each intrinsic mode function; based on the optimal denoising parameters corresponding to the intrinsic mode functions, the corresponding intrinsic mode functions are subjected to denoising processing and then combined, and firefighting pipe network monitoring data after denoising are obtained; according to the method, the fire-fighting pipe network monitoring data can be denoised, and the effectiveness of data feature extraction is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of fire water supply network status monitoring and fault diagnosis, and in particular to a fire water supply network monitoring data denoising method and system. Background Art

[0002] Converter stations are crucial power transmission and conversion facilities. Ensuring the proper operation of firefighting networks is crucial for timely extinguishing fires. Typically, pressure / flow sensors deployed on these networks collect real-time pressure / flow data. Data analysis and feature extraction are then performed to further monitor the firefighting water supply network and diagnose faults. However, due to harsh operating environments, the pressure / flow data collected by these sensors is susceptible to environmental influences, such as pipe vibration caused by wind. Therefore, directly using raw pressure / flow data for data analysis and feature extraction often results in inaccurate results that fail to reflect the true operational status of the firefighting network. Therefore, prior to data analysis and feature extraction, denoising the raw pressure / flow data is essential. Due to the complexity of environmental noise, traditional data denoising methods often have significant limitations. For example, they are only effective at denoising high-frequency or low-frequency noise, failing to fully capture the noise distribution of the pressure / flow data.

[0003] In related technologies, the document “Research on Flow Verification and Leakage Detection of Water Supply Network Nodes, Zhang Jing, Master’s Thesis of Kunming University of Science and Technology” proposes to combine wavelet threshold and Kalman filtering to denoise the monitoring data of water supply network, so as to improve the robustness of the water supply network leakage detection algorithm; however, the wavelet threshold denoising method used in this document often requires prior knowledge of the frequency distribution of the noise, while in reality, noise is usually randomly distributed, with a large distribution range and is not concentrated. The adaptability of wavelet denoising is poor, and the real components of the signal are destroyed while denoising; the Kalman filtering algorithm is highly complex, requires prior knowledge of the system model, and has high requirements on the quality of the observation values, making it difficult to meet denoising scenarios with high timeliness requirements and low observation quality. Summary of the Invention

[0004] The technical problem to be solved by the present invention is how to achieve denoising of fire pipe network monitoring data and improve the effectiveness of data feature extraction.

[0005] The present invention solves the above technical problems through the following technical means:

[0006] A fire protection pipe network monitoring data denoising method is proposed, which includes:

[0007] Perform variational mode decomposition on the fire protection pipe network monitoring data to obtain multiple eigenmode functions;

[0008] For each of the intrinsic mode functions, optimizing the parameters of the non-local means algorithm to obtain the optimal denoising parameters corresponding to each of the intrinsic mode functions;

[0009] Based on the optimal denoising parameters corresponding to each of the eigenmode functions, the corresponding eigenmode functions are denoised and then merged to obtain denoised fire pipe network monitoring data.

[0010] Furthermore, the constraint model of the variational mode decomposition is:

[0011]

[0012] Where: u k (t)={u1(t),u2(t),…,u j (t)} is each of the eigenmode functions; ω k ={ω1,ω2,…,ω j} is the center frequency of each of the eigenmode functions; e is the base of the natural logarithm; j is the imaginary unit; t is time; f(t) is the original signal, is the partial derivative with respect to time t, and ‖‖ is the Euclid norm of each modal component.

[0013] Furthermore, for each of the intrinsic mode functions, the parameters of the non-local means algorithm are optimized to obtain the optimal denoising parameters corresponding to each of the intrinsic mode functions, including:

[0014] For each of the intrinsic mode functions, denoising the intrinsic mode function using a non-local means algorithm based on the initialized denoising parameters, and calculating the error between a theoretical estimate of the true signal after denoising and the true signal;

[0015] Add the denoising parameters and errors corresponding to this iterative process to the observed data set;

[0016] The optimal value of the observed data set obtained after n iterations is estimated to obtain the optimal denoising parameter corresponding to the current intrinsic mode function. The formula is expressed as:

[0017]

[0018] Where z * To obtain the optimal denoising parameter that minimizes the objective function ∈(z), R d is a d-dimensional real space, and the observation data set (z,∈(z)) is {(z1,∈(z1)),(z2,∈(z2)),…,(z n ,∈(z n ))}, z1, z2, …z nare the denoising parameters of the non-local mean algorithm in each iteration process; ∈(z1), ∈(z2)…, ∈(z n ) is the error between the theoretical estimate of the true signal after denoising and the true signal in each iteration.

[0019] Furthermore, the denoising parameters of the non-local means algorithm include a search area half-width K, a similar structure block half-width P, and a bandwidth parameter λ.

[0020] Furthermore, the non-local means algorithm is used to perform denoising on the intrinsic mode function, and the error between the theoretical estimate of the real signal after denoising and the real signal is calculated, including:

[0021] The non-local mean algorithm is used to denoise the intrinsic mode function and obtain the theoretical estimate of the true signal after denoising. The formula is expressed as:

[0022]

[0023] Where, is the theoretical estimate of the true signal after denoising; Z(s) = ∑ t∈D(s) ω(s, t) is the normalization factor, D(s) is the search area, ω(s, t) is the similarity between a similar structure block in the search area and the target structure block; u k (t) is the eigenmode function;

[0024] Calculate the error between the theoretical estimate of the true signal after denoising and the true signal.

[0025] Furthermore, the calculation formula of the similarity degree ω(s, t) is:

[0026]

[0027] Where: Δ is a similar structural block centered on s; L Δ is the target structure block centered at t; d 2 (s, t) is the sum of the squares of the Euclidean distances between points s and t; λ is the bandwidth parameter.

[0028] Furthermore, after denoising the corresponding intrinsic mode functions based on the optimal denoising parameters corresponding to the respective intrinsic mode functions and then merging them to obtain denoised fire pipe network monitoring data, the method further includes:

[0029] Calculate the root mean square error and signal-to-noise ratio of the denoised fire pipe network monitoring data;

[0030] Based on the root mean square error and signal-to-noise ratio, the denoised fire pipe network monitoring data is evaluated.

[0031] Furthermore, after evaluating the denoised fire pipe network monitoring data based on the root mean square error and the signal-to-noise ratio, the method further includes:

[0032] After the evaluation results are passed, data analysis and feature extraction are performed based on the denoised fire pipe network monitoring data.

[0033] Furthermore, the fire protection pipe network monitoring data includes pressure data and / or flow data of the pipe network.

[0034] In addition, the present invention also proposes a fire protection pipe network monitoring data denoising system, the system comprising:

[0035] The modal decomposition module is used to perform variational modal decomposition on the fire pipe network monitoring data to obtain multiple intrinsic mode functions;

[0036] A parameter optimization module, configured to optimize the parameters of the non-local means algorithm for each of the intrinsic mode functions to obtain the optimal denoising parameters corresponding to each of the intrinsic mode functions;

[0037] The monitoring data denoising module is used to denoise the corresponding intrinsic mode functions based on the optimal denoising parameters corresponding to each of the intrinsic mode functions and then merge them to obtain denoised fire pipe network monitoring data.

[0038] The advantages of the present invention are:

[0039] The present invention first adopts variational mode decomposition (VMD) to solve the variational model, determines the center frequency and bandwidth of each mode, decomposes the original monitoring data into multiple eigenmode functions with limited bandwidth, and further denoises the eigenmode functions with different bandwidths. Since the denoising effect of the non-local means (NLM) algorithm is greatly affected by the parameters, the present invention further adopts Bayesian optimization (BO) to optimize the NLM algorithm parameters to obtain the optimal denoising parameters corresponding to each of the eigenmode functions, and uses the optimal denoising parameters to denoise the corresponding eigenmode functions and then merge them to obtain denoised fire pipe network monitoring data. Compared with the traditional one-dimensional monitoring data denoising method, this method avoids the limitations of human experience, improves the adaptability of the NLM algorithm, reduces the adverse effects of environmental noise on data analysis, and improves the effectiveness of data feature extraction.

[0040] Additional aspects and advantages of the present invention will be set forth in part in the description which follows and, in part, will be obvious from the description which follows, or may be learned through practice of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0041] Figure 1This is a flow chart of a method for denoising fire pipe network monitoring data according to an embodiment of the present invention;

[0042] Figure 2 is a flow chart of the BO-NLM algorithm in one embodiment of the present invention;

[0043] Figure 3 is a real signal excluding environmental noise or a signal including environmental noise in one embodiment of the present invention;

[0044] Figure 4 2 is a schematic diagram of a result of VMD decomposition of a signal containing environmental noise according to an embodiment of the present invention;

[0045] Figure 5 1 is a schematic diagram of the denoising results of an intrinsic mode function using an NLM algorithm with optimized parameters in one embodiment of the present invention;

[0046] Figure 6 1. It is a schematic diagram showing the result of denoising a signal containing environmental noise using the BO-NLM algorithm in one embodiment of the present invention;

[0047] Figure 7 Schematic diagram of a fire protection pipe network monitoring data denoising system proposed in one embodiment of the present invention. DETAILED DESCRIPTION

[0048] To make the objectives, technical solutions, and advantages of the embodiments of the present invention more clear, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. All other embodiments obtained by ordinary technicians in this field based on the embodiments of the present invention without making any creative efforts shall fall within the scope of protection of the present invention.

[0049] like Figure 1 As shown, an embodiment of the present invention provides a method for denoising fire pipe network monitoring data, the method comprising the following steps:

[0050] S10, performing variational mode decomposition on the fire protection pipe network monitoring data to obtain multiple eigenmode functions;

[0051] It should be noted that, in this embodiment, the monitoring data of the fire protection pipe network is collected through pressure sensors and flow sensors arranged on the fire protection pipe network, and the monitoring data includes pressure data and flow data.

[0052] It should be noted that, before denoising the original fire protection pipe network monitoring data, this embodiment uses variational modal decomposition to decompose the original monitoring data. By iteratively solving the variational model, the center frequency and bandwidth of each mode are determined, and it is decomposed into multiple finite bandwidth intrinsic mode functions.

[0053] S20, optimizing the parameters of the non-local means algorithm for each of the intrinsic mode functions to obtain the optimal denoising parameters corresponding to each of the intrinsic mode functions;

[0054] S30 , performing denoising processing on the corresponding intrinsic mode functions based on the optimal denoising parameters corresponding to the respective intrinsic mode functions, and then merging the denoised fire pipe network monitoring data to obtain denoised fire pipe network monitoring data.

[0055] The non-local means denoising algorithm employed in this embodiment utilizes the redundant information commonly found in signals for denoising. It exploits the overall characteristics of the signal and effectively removes Gaussian noise, a predominant form of noise in nature. The non-local means denoising algorithm does not require prior knowledge of the noise frequency distribution or system model, nor does it place high demands on the quality of the observations. By utilizing the inherent characteristic information of the signal for denoising, it is highly adaptable to the signal and effectively preserves the true components of the original signal.

[0056] As a further preferred technical solution, the constraint model of the variational mode decomposition is:

[0057]

[0058] Where: u k (t)={u1(t),u2(t),…,u j (t)} is each of the eigenmode functions; ω k ={ω1,ω2,…,ω j} is the center frequency of each of the eigenmode functions; e is the base of the natural logarithm; j is the imaginary unit; t is time; f(t) is the original signal, is the partial derivative with respect to time t, and ‖‖ is the Euclid norm of each modal component.

[0059] It should be noted that VMD is an adaptive signal decomposition method with a rigorous theoretical foundation that can effectively overcome the endpoint effects and modal aliasing effects of traditional signal decomposition methods. Assume that the original monitoring signal f(t) is decomposed into k components, ensuring that the decomposition sequence is a modal component with a finite bandwidth and a center frequency. The objective function is to minimize the sum of the estimated bandwidths of each mode, with the constraint that the sum of each mode is equal to the original signal.

[0060] As a further preferred technical solution, step S20: for each of the intrinsic mode functions, optimizing the parameters of the non-local means algorithm to obtain the optimal denoising parameters corresponding to each of the intrinsic mode functions, specifically includes the following steps:

[0061] S21. For each of the intrinsic mode functions, denoising the intrinsic mode function using a non-local means algorithm based on the initialized denoising parameters, and calculating the error between the theoretical estimate of the true signal after denoising and the true signal;

[0062] S22, adding the denoising parameters and errors corresponding to this iterative process to the observation data set;

[0063] S23, perform optimal value estimation on the observation data set obtained after n iterations to obtain the optimal denoising parameter corresponding to the current intrinsic mode function, which is expressed as follows:

[0064]

[0065] Where z * To obtain the optimal denoising parameter that minimizes the objective function ∈(z), R d is a d-dimensional real space, and the observation data set (z,∈(z)) is {(z1,∈(z1)),(z2,∈(z2)),…,(z n ,∈(z n ))}, z1, z2, …z n are the denoising parameters of the non-local mean algorithm in each iteration process; ∈(z1), ∈(z2)…, ∈(z n ) is the error between the theoretical estimate of the true signal after denoising and the true signal in each iteration.

[0066] Specifically, since the original monitoring data consists of the real pressure / flow signal x and the noise signal x * Composition, including environmental noise, namely:

[0067] y(t)=x(t)+x * (t)

[0068] Where: y(t) is the original monitoring data collected; x(t) is the real monitoring data without environmental noise; x * (t) is the environmental noise signal.

[0069] The decomposed IMFs also contain ambient noise. Furthermore, different IMFs have different center frequencies, and therefore contain ambient noise at different frequencies. Using the same set of parameters to denoise all IMFs clearly makes it difficult to fully account for the noise distribution at different frequencies. Therefore, denoising each IMF separately is necessary.

[0070] The denoising effect of the NLM algorithm is greatly affected by parameters, including the search area half-width K, the similar structure block half-width P, and the bandwidth parameter λ. If the search area half-width K is too large, it is easy to ignore key information, and if it is too small, artifacts are likely to appear. If the similar structure block half-width P is too large, it is difficult to find similar structure blocks, and if it is too small, too many similar structure blocks will be found. The bandwidth parameter λ is a parameter that controls the smoothness of the denoised signal. If λ is too small, it is easy to cause noise fluctuations and poor denoising effect. If it is too large, the signal will be blurred and a large amount of detailed information will be lost. Therefore, to ensure the denoising effect, this embodiment needs to optimize the NLM algorithm parameters and perform denoising processing on each intrinsic mode function separately.

[0071] The Bayesian Optimization (BO) algorithm has a low requirement for the number of sampling points and has the advantages of a small number of iterations and high speed. Therefore, the present invention uses Bayesian Optimization (BO) to optimize the parameters of the NLM algorithm. The BO algorithm consists of two parts, namely a Gaussian process and an extraction function. After obtaining the posterior probability of the function through the Gaussian process, the extraction function samples new points based on certain indicators of the posterior probability. The sampled new points are added to the observation data as a reference for the next calculation, which can describe a more accurate posterior probability.

[0072] In the present invention, z = {K, P, λ} is the NLM algorithm parameter, that is, the non-local means algorithm denoising parameters include the search area half-width K, the similar structure block half-width P and the bandwidth parameter λ. ∈(z) is the error between the theoretical estimate of the real signal after denoising and the real signal, that is Since the denoising effect of the NLM algorithm is greatly affected by the parameters, different algorithm parameters z have different values of the error function ∈(z). The goal of the BO algorithm is to optimize the algorithm parameters z. * , so that∈(z * ) takes the minimum value. Assume that after the nth iteration, the algorithm parameters and error are (z n ,∈(z n )). Before the n+1th iteration, add the nth iteration value to the known observation data {(z1,∈(z1)),(z2,∈(z2)),…,(z n ,∈(z n ))}, which obey the Gaussian distribution G(μ(z),σ 2 (z)), μ(z) is the mean, σ 2 (z) is the variance, and then the n+1th iteration value is obtained by the extraction function, that is:

[0073] z n+1 =argmaxμ(z)+β 1 / 2 σ 2 (z)

[0074] Where: z n+1is the n+1th iteration value, and β is the BO algorithm parameter. After continuous iteration, the parameter combination z that makes ∈(z) take the minimum value is selected. * =(K * ,P * ,λ * ).

[0075] It should be noted that the purpose of iteration is to continuously update the NLM algorithm parameters. With each iteration, the NLM algorithm parameters are optimized, resulting in a better denoising effect for the NLM algorithm. This embodiment iteratively updates the NLM algorithm parameters, selects the optimal algorithm parameters, introduces them into the NLM algorithm, and uses this set of optimal parameters to denoise the original signal. Denoising the original signal only requires one iteration to optimize the appropriate algorithm parameters, which are then used for denoising.

[0076] like Figure 2 As shown, this embodiment uses the BO algorithm to optimize the corresponding search area half-width K, similar structure block half-width P and bandwidth parameter λ for the intrinsic mode functions of different frequencies of the decomposed pressure / flow data, avoiding the limitation of human experience and improving the adaptability of the NLM algorithm.

[0077] As a further preferred technical solution, in step S21, the non-local means algorithm is used to denoise the intrinsic mode function, and the error between the theoretical estimate of the real signal after denoising and the real signal is calculated, including:

[0078] The non-local mean algorithm is used to denoise the intrinsic mode function and obtain the theoretical estimate of the true signal after denoising. The formula is expressed as:

[0079]

[0080] Where, is the theoretical estimate of the true signal after denoising; z(s) = ∑ t∈D(s) ω(s, t) is the normalization factor, D(s) is the search area, ω(s, t) is the similarity between a similar structure block in the search area and the target structure block; u k (t) is the eigenmode function;

[0081] Calculate the error between the theoretical estimate of the true signal after denoising and the true signal.

[0082] As a further preferred technical solution, the similarity degree ω(s, t) depends on the physical distance between the center point s of a similar structure block and the center point t of the target structure block, and is calculated as follows:

[0083]

[0084] Where: Δ is a similar structural block centered on s; L Δ is the target structure block centered at t; d 2 (s, t) is the sum of the squares of the Euclidean distances between points s and t; λ is the bandwidth parameter.

[0085] As a further preferred technical solution, after the step S30 of denoising the corresponding intrinsic mode functions based on the optimal denoising parameters corresponding to the respective intrinsic mode functions and then merging them to obtain denoised fire pipe network monitoring data, the method further includes:

[0086] Calculate the root mean square error and signal-to-noise ratio of the denoised fire pipe network monitoring data;

[0087] Based on the root mean square error and signal-to-noise ratio, the denoised fire pipe network monitoring data is evaluated.

[0088] Specifically, the eigenmode functions of the denoised pressure / flow data are combined to give a theoretical estimate of the true denoised signal. To quantitatively evaluate the denoising effect, the root mean square error RMSE and signal-to-noise ratio SNR are defined, that is,

[0089]

[0090] Where: N is the number of sampling points, x(t) is the real signal without environmental noise, is the theoretical estimate of the true signal after denoising. Obviously, the smaller the RMSE and the larger the SNR, the better the denoising effect of the algorithm.

[0091] As a further preferred technical solution, after evaluating the denoised fire pipe network monitoring data based on the root mean square error and signal-to-noise ratio, the method further includes:

[0092] After the evaluation results are passed, data analysis and feature extraction are performed based on the denoised fire pipe network monitoring data.

[0093] The following is an example of a specific application:

[0094] (1) The pressure and flow data of the fire protection pipe network are collected in real time through the pressure sensors and flow sensors on the fire protection pipe network as monitoring data. Figure 3 As shown in the figure, the monitoring data of the simulated fire pipe network is constructed as follows:

[0095]

[0096] Where: A i＝1,2,…,6 are amplitudes, which are 20, 4.5, 2.55, 1.5, 0.4 and 0.3 respectively, f i＝1,2,…,6are frequencies, which are 1.25, 1.25×2, 1.25×3, 1.25×4, 1.25×0.2 and 1.25×0.3 respectively. To simulate the influence of environmental noise, a Gaussian white noise x with a signal-to-noise ratio of 5dB is superimposed on the real signal x(t) without environmental noise. * (t), that is, the collected pressure / flow signal is y(t)=x(t)+x * (t).

[0097] (2) Variational mode decomposition (VMD) is used to decompose the original monitoring data. By iteratively solving the variational model, the center frequency and bandwidth of each mode are determined and decomposed into multiple finite bandwidth intrinsic mode functions, such as Figure 4 As shown in Figure 2, it is clear that different eigenmode functions have different frequencies.

[0098] (3) Different eigenmode functions have different center frequencies, and the frequencies of the ambient noise they contain are also different. If the same set of parameters is used to denoise different eigenmode functions, it is difficult to fully cover the noise distribution of different frequencies. Therefore, it is necessary to denoise each eigenmode function separately.

[0099] The BO algorithm is used to optimize the NLM algorithm parameters, and the NLM algorithm parameters for different intrinsic mode functions are determined as shown in Table 1. Since the frequencies of the ambient noise contained in different intrinsic mode functions are different, their corresponding NLM algorithm optimization parameters are also different.

[0100] Table 1

[0101]

[0102] (4) According to the different intrinsic mode functions of the monitoring data, the NLM algorithm with the corresponding optimized parameters is used to denoise the intrinsic mode functions of the monitoring data, such as Figure 5 shown.

[0103] (5) Combine the intrinsic mode functions of the denoised pressure / flow data and the theoretical estimate of the true signal after denoising To quantitatively evaluate the denoising effect, the root mean square error RMSE and signal-to-noise ratio SNR are defined, that is,

[0104]

[0105] Where: N is the number of sampling points, x(t) is the real signal without environmental noise, is the theoretical estimate of the true signal after denoising. Figure 6 As shown in the figure, the NLM algorithm with optimized parameters has a very significant denoising effect on the signal containing environmental noise. The denoised signal has RMSE = 0.1314 and SNR = 40.9635.

[0106] (6) Perform data analysis and feature extraction based on the denoised monitoring data.

[0107] In addition, if Figure 7 As shown, another embodiment of the present invention further provides a fire protection pipe network monitoring data denoising system, the system comprising:

[0108] The modal decomposition module 10 is used to perform variational modal decomposition on the fire pipe network monitoring data to obtain multiple intrinsic mode functions;

[0109] A parameter optimization module 20 is used to optimize the parameters of the non-local means algorithm for each of the intrinsic mode functions to obtain the optimal denoising parameters corresponding to each of the intrinsic mode functions;

[0110] The monitoring data denoising module 30 is used to denoise the corresponding intrinsic mode functions based on the optimal denoising parameters corresponding to each of the eigenmode functions and then merge them to obtain the denoised fire pipe network monitoring data.

[0111] As a further preferred technical solution, the parameter optimization module 20 includes:

[0112] an error calculation unit, configured to perform denoising processing on each of the intrinsic mode functions using a non-local means algorithm based on the initialized denoising parameters, and calculate the error between a theoretical estimate of a true signal after denoising and the true signal;

[0113] An observation data set construction unit is used to add the denoising parameters and errors corresponding to this iterative process into the observation data set;

[0114] The iteration unit is used to estimate the optimal value of the observation data set obtained after n iterations to obtain the optimal denoising parameter corresponding to the current intrinsic mode function. The formula is expressed as:

[0115]

[0116] Where z * To obtain the optimal denoising parameter that minimizes the objective function ∈(z), R d is a d-dimensional real space, and the observation data set (z,∈(z)) is {(z1,∈(∈1)),(z2,∈(z2)),…,(z n ,∈(z n ))}, z1, z2, …z n are the denoising parameters of the non-local mean algorithm in each iteration process; ∈(z1), ∈(z2)…, ∈(z n ) is the error between the theoretical estimate of the true signal after denoising and the true signal in each iteration.

[0117] As a further preferred technical solution, the error calculation unit is specifically used to:

[0118] The non-local mean algorithm is used to denoise the intrinsic mode function and obtain the theoretical estimate of the true signal after denoising. The formula is expressed as:

[0119]

[0120] Where, is the theoretical estimate of the true signal after denoising; z(s) = ∑ t∈D(s) ω(s, t) is the normalization factor, D(s) is the search area, ω(s, t) is the similarity between a similar structure block in the search area and the target structure block; u k (t) is the eigenmode function;

[0121] Calculate the error between the theoretical estimate of the true signal after denoising and the true signal.

[0122] As a further preferred technical solution, the calculation formula of the similarity degree ω(s, t) is:

[0123]

[0124] Where: Δ is a similar structural block centered on s; L Δ is the target structure block centered at t; d 2 (s, t) is the sum of the squares of the Euclidean distances between points s and t; λ is the bandwidth parameter.

[0125] It should be noted that other embodiments or specific implementation methods of the fire pipe network monitoring data denoising system of the present invention can refer to the above-mentioned method embodiments and will not be repeated here.

[0126] Throughout this specification, reference to terms such as "one embodiment," "some embodiments," "examples," "specific examples," or "some examples" means that a specific feature, structure, material, or characteristic described in conjunction with that embodiment or example is included in at least one embodiment or example of the present invention. In this specification, schematic representations of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in any one or more embodiments or examples.

[0127] Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of the technical features being referred to. Thus, a feature defined as "first" or "second" may explicitly or implicitly include at least one such feature. In the description of the present invention, "plurality" means at least two, such as two, three, etc., unless otherwise specifically defined.

[0128] Although the embodiments of the present invention have been shown and described above, it will be understood that the above embodiments are illustrative and are not to be construed as limitations on the present invention. A person skilled in the art may change, modify, replace and modify the above embodiments within the scope of the present invention.

Claims

1. A method for denoising fire pipe network monitoring data, characterized in that: The method comprises: Perform variational mode decomposition on the fire protection pipe network monitoring data to obtain multiple eigenmode functions; For each of the intrinsic mode functions, optimizing the parameters of the non-local means algorithm to obtain the optimal denoising parameters corresponding to each of the intrinsic mode functions; Based on the optimal denoising parameters corresponding to each of the eigenmode functions, the corresponding eigenmode functions are denoised and then merged to obtain denoised fire pipe network monitoring data.

2. The fire protection pipe network monitoring data denoising method according to claim 1, characterized in that: The constraint model of the variational mode decomposition is: Where: u k (t)={u1(t),u2(t),…,u j (t)} is each of the eigenmode functions; ω k ={ω1,ω2,…,ω j } is the center frequency of each of the eigenmode functions; e is the base of the natural logarithm; j is the imaginary unit; t is time; f(t) is the original signal, is the partial derivative with respect to time t, and ‖‖ is the Euclid norm of each modal component.

3. The fire protection pipe network monitoring data denoising method according to claim 1, characterized in that: For each of the intrinsic mode functions, the parameters of the non-local means algorithm are optimized to obtain the optimal denoising parameters corresponding to each of the intrinsic mode functions, including: For each of the intrinsic mode functions, denoising the intrinsic mode function using a non-local means algorithm based on the initialized denoising parameters, and calculating the error between a theoretical estimate of the true signal after denoising and the true signal; Add the denoising parameters and errors corresponding to this iterative process to the observed data set; The optimal value of the observed data set obtained after n iterations is estimated to obtain the optimal denoising parameter corresponding to the current intrinsic mode function. The formula is expressed as: Where z * To obtain the optimal denoising parameter that minimizes the objective function ∈(z), R d is a d-dimensional real space, and the observation data set (z,∈(z)) is {(z1,∈(z1)),(z2,∈(z2)),…,(z n ,∈(z n ))}, z1, z2, …z n are the denoising parameters of the non-local mean algorithm in each iteration process; ∈(z1), ω(z2)…, ∈(z n ) is the error between the theoretical estimate of the true signal after denoising and the true signal in each iteration.

4. The fire protection pipe network monitoring data denoising method according to claim 3, characterized in that: The denoising parameters of the non-local means algorithm include the search area half-width K, the similar structure block half-width P and the bandwidth parameter λ.

5. The fire protection pipe network monitoring data denoising method according to claim 3, characterized in that: The non-local mean algorithm is used to perform denoising on the intrinsic mode function, and the error between the theoretical estimated value of the real signal after denoising and the real signal is calculated, including: The non-local mean algorithm is used to denoise the intrinsic mode function and obtain the theoretical estimate of the true signal after denoising. The formula is expressed as: Where, is the theoretical estimate of the true signal after denoising; Z(s) = ∑ t∈D(s) ω(s, t) is the normalization factor, D(s) is the search area, ω(s, t) is the similarity between a similar structure block in the search area and the target structure block; u k (t) is the eigenmode function; Calculate the error between the theoretical estimate of the true signal after denoising and the true signal.

6. The fire protection pipe network monitoring data denoising method according to claim 1, characterized in that: The calculation formula of the similarity degree ω(s,t) is: Where: Δ is a similar structural block centered on s; L Δ is the target structure block centered at t; d 2 (s, t) is the sum of the squares of the Euclidean distances between points s and t; λ is the bandwidth parameter.

7. The fire protection pipe network monitoring data denoising method according to claim 1, characterized in that: After denoising the corresponding intrinsic mode functions based on the optimal denoising parameters corresponding to the respective intrinsic mode functions and then merging them to obtain denoised fire pipe network monitoring data, the method further includes: Calculate the root mean square error and signal-to-noise ratio of the denoised fire pipe network monitoring data; Based on the root mean square error and signal-to-noise ratio, the denoised fire pipe network monitoring data is evaluated.

8. The fire protection pipe network monitoring data denoising method according to claim 1, characterized in that: After evaluating the denoised fire pipe network monitoring data based on the root mean square error and the signal-to-noise ratio, the method further includes: After the evaluation results are passed, data analysis and feature extraction are performed based on the denoised fire pipe network monitoring data.

9. The fire protection pipe network monitoring data denoising method according to any one of claims 1 to 8, characterized in that: The fire protection pipe network monitoring data includes pressure data and / or flow data of the pipe network.

10. A fire protection pipe network monitoring data denoising system, characterized in that: The system comprises: The modal decomposition module is used to perform variational modal decomposition on the fire pipe network monitoring data to obtain multiple intrinsic mode functions; A parameter optimization module, configured to optimize the parameters of the non-local means algorithm for each of the intrinsic mode functions to obtain the optimal denoising parameters corresponding to each of the intrinsic mode functions; The monitoring data denoising module is used to denoise the corresponding intrinsic mode functions based on the optimal denoising parameters corresponding to each of the intrinsic mode functions and then merge them to obtain denoised fire pipe network monitoring data.

Images