An ISSA-HC-based PCCP Broken Wire Signal Analysis System

Abstract

The present invention discloses a PCCP broken wire signal analysis system based on ISSA-HC, which relates to the technical fields of distributed optical fiber sensing and artificial intelligence. The present invention includes a Φ-OTDR type signal monitoring module, a feature calculation module, a feature optimization module, a vibration mode recognition module, and a visualization display module that are connected in sequence. Among them, the Φ-OTDR type signal monitoring module is used to collect vibration signals during the operation of PCCP, the feature calculation module is used to calculate the first-level feature parameters of the vibration signals, the feature optimization module is used to extract effective feature parameters from the first-level feature parameters, the vibration mode recognition module is used to identify the types of vibration signals based on the set of effective feature parameters, and the visualization display module is used to intuitively present the results to the operators. The purpose of the present invention is to improve the accuracy of real-time monitoring of PCCP broken wire signals and provide technical support for the health diagnosis and disaster warning of in-service PCCP projects.

Description

Technical Field

[0001] The present invention relates to the technical fields of distributed optical fiber sensing and artificial intelligence, and particularly relates to a PCCP broken wire signal analysis system based on ISSA-HC. Background Art

[0002] Prestressed concrete cylinder pipe (PCCP) is a composite pipe composed of a concrete pipe core, a steel cylinder, prestressed steel wires, and a mortar protective layer. Due to its advantages such as high strength, strong impermeability, good corrosion resistance, and long service life, it has been widely used in long-distance water conveyance projects. The bearing capacity of PCCP depends on the prestressed steel wires wound around the pipe core. Affected by the corrosion environment or under the action of hydrogen embrittlement effect, the prestressed steel wires may break. When the number of broken wires reaches a certain amount, there is a great risk of pipe explosion, and the consequences are very serious, not only causing great casualties and property losses, but also causing serious secondary disasters. Adopting scientific real-time monitoring means helps technicians accurately obtain information such as the time, location, and quantity of PCCP broken wires, which is of great significance for ensuring the safe service of the project.

[0003] Distributed optical fiber sensing is an advanced PCCP broken wire monitoring technology, which has many advantages such as long monitoring distance, high spatio-temporal resolution, fast response speed, and intrinsic safety. It has been successfully applied to practical projects and achieved remarkable results. Phase-sensitive optical time domain reflectometry (Φ-OTDR) is a new type of distributed optical fiber monitoring technology, which can be divided into direct detection type and coherent detection type in terms of structure. Compared with the direct detection type, the coherent detection type has a wider dynamic range and higher signal-to-noise ratio, and is suitable for long-distance detection of weak signals. To effectively detect PCCP broken wires, it is not only necessary to optimize and improve the signal monitoring system, but also necessary to develop an efficient signal analysis system to scientifically extract the vibration characteristics of the monitoring signal, and then accurately identify the type of abnormal signal. The effectiveness of feature parameters has an important impact on ensuring the accuracy of pattern recognition. At present, some scholars have constructed various types of feature parameters from the perspectives of time domain, frequency domain, time-frequency domain, space domain, etc. However, some feature parameters are invalid or redundant. If the effectiveness of feature parameters is ignored during pattern recognition, it will not only increase the computational complexity, but also cause the decline of identification performance. Therefore, it is necessary to establish scientific theories and algorithms to analyze and extract the set of parameters that can effectively reflect the characteristics of PCCP broken wire signals. The wrapper theory is a basic effective feature extraction mode, which integrates feature parameter search and signal classification, and uses classification performance as the standard to measure the quality of feature subsets, including four parts: feature subset search, feature subset evaluation, stopping criterion, and result verification. Although the wrapper algorithm has a high degree of complexity and is prone to overfitting, the extracted feature subset generally has good pattern recognition performance.

[0004] The core of the encapsulated algorithm lies in constructing a scientific feature subset search strategy. Currently, three basic modes have been formed, namely: global search, sequential search, and random search. Each of the three modes has its own advantages, disadvantages, and applicable conditions. The advantage of global search is that it can find the global optimal solution, but the time complexity is too high, and it is applicable to the case where the dimension of the feature parameter set is relatively low; the time complexity of sequential search is the lowest, but only a local optimal solution can be found, and the pattern recognition performance is poor; random search combines the advantages of global search and sequential search. Its time complexity is lower than that of global search but higher than that of sequential search, and it can find an approximate global optimal solution, which is applicable to solving the technical problem of effective feature mining of PCCP broken wire signals. The Sparrow Search Algorithm (SSA) is a meta-heuristic algorithm established based on the foraging behavior and predator avoidance behavior of the sparrow population. In SSA, the sparrow population is divided into discoverers, followers, and scouts. The discoverers are responsible for guiding the population to forage, the followers follow the discoverers to forage, and the scouts undertake the functions of reconnaissance and early warning. The standard SSA has strong global exploration and local development capabilities, but when solving high-dimensional and complex optimization problems, its learning ability is still insufficient. Existing achievements focus on improving the search ability of SSA in the neighborhood space, without fundamentally changing the optimization mechanism of the algorithm. Especially when the search approaches the global optimal solution, the population diversity will be significantly reduced, and it is easy to fall into the local optimum. To improve the search performance of SSA, the improvement strategy of the algorithm still needs to be studied in depth. Hierarchical Clustering (HC) belongs to the category of unsupervised learning. Without prior knowledge, it classifies research objects according to the similarity principle. The classification accuracy of HC for vibration signals can provide an effective way for feature subset evaluation.

[0005] In summary, comprehensively applying modern mathematics, distributed optical fiber sensing, and artificial intelligence technologies to develop a PCCP broken wire signal analysis system based on ISSA-HC to solve the difficulties existing in the prior art is a bottleneck problem that those skilled in the art urgently need to solve. Summary of the Invention

[0006] In view of this, the present invention discloses a PCCP broken wire signal analysis system based on ISSA-HC, aiming to improve the accuracy of real-time monitoring of PCCP broken wire signals and provide technical support for the health diagnosis and disaster early warning of in-service PCCP projects.

[0007] To achieve the above object, the present invention adopts the following technical solutions.

[0008] A PCCP broken wire signal analysis system based on ISSA-HC includes a Φ-OTDR type signal monitoring module, a feature calculation module, a feature optimization module, a vibration pattern recognition module, and a visualization display module connected in sequence;

[0009] Among them, the Φ-OTDR type signal monitoring module is used to collect the vibration signals of PCCP during operation;

[0010] The feature calculation module is used to perform time-domain, frequency-domain, and time-frequency-domain analyses on the collected vibration signals, and calculate the first-level feature parameters;

[0011] The feature optimization module is used to extract effective feature parameters from the first-level feature parameters;

[0012] The vibration mode recognition module identifies the type of vibration signal based on the set of effective feature parameters;

[0013] The visualization display module is used to intuitively present the feature analysis and pattern recognition results to the operator.

[0014] For the above system, optionally, the Φ-OTDR type signal monitoring module is based on a coherent detection design, and includes: a laser, a first beam splitter, an acousto-optic modulator, an erbium-doped fiber amplifier, a circulator, a second beam splitter, a photodetector, and a data acquisition card;

[0015] Among them, the output end of the laser is connected to the input end of the first beam splitter, the first output end of the first beam splitter is connected to the input end of the acousto-optic modulator, the output end of the acousto-optic modulator is connected to the input end of the erbium-doped fiber amplifier, the output end of the erbium-doped fiber amplifier is connected to the first port of the circulator, the sensing fiber is arranged closely along the inner wall of the PCCP, the output end of the sensing fiber is connected to the second port of the circulator, the third port of the circulator is connected to the first input end of the second beam splitter, the second output end of the first beam splitter is connected to the second input end of the second beam splitter, the output end of the second beam splitter is connected to the input end of the photodetector, and the output end of the photodetector is connected to the input end of the data acquisition card.

[0016] For the above system, optionally, the feature calculation module is used to analyze the data in the data acquisition card and calculate the first-level feature parameters, including a time-domain calculation unit, a frequency-domain calculation unit, and a time-frequency-domain calculation unit connected in sequence.

[0017] For the above system, optionally, the first-level feature parameters include: time-domain feature parameters, frequency-domain feature parameters, and time-frequency-domain feature parameters;

[0018] Among them, the time-domain calculation unit calculates the time-domain feature parameters by analyzing the time-domain signal, the frequency-domain calculation unit maps the time-domain signal to the frequency domain using the discrete Fourier transform, and then calculates the frequency-domain feature parameters, and the time-frequency-domain calculation unit performs multi-frequency decomposition on the time-domain signal based on the discrete wavelet transform, and accordingly calculates the time-frequency-domain feature parameters.

[0019] For the above system, optionally, the feature optimization module includes a filtering optimization unit and a packaging optimization unit connected in sequence;

[0020] Among them, the output end of the filtering optimization unit is connected to the input end of the encapsulation optimization unit.

[0021] For the above system, optionally, in the filtering optimization unit, the first-level feature parameters output by the feature calculation module are judged according to the Fisher criterion, and the second-level feature parameters that meet the judgment criterion are selected.

[0022] For the above system, optionally, in the encapsulation optimization unit, by improving the SSA, an ISSA optimization strategy is established, and the classification accuracy of the vibration signal by the HC is used as the discrimination criterion for evaluating the quality of the feature subset. On this basis, an ISSA-HC integration framework is constructed by means of the encapsulation modeling theory, and the second-level feature parameters output by the filtering optimization unit are deeply mined according to the ISSA-HC, and an effective feature parameter set is established.

[0023] For the above system, optionally, the improvement process of the SSA is as follows: an adaptive adjustment strategy is introduced to enhance the global search ability of the algorithm in the early stage and the local search ability in the later stage, and the Gaussian mutation perturbation and the Tent chaotic perturbation are alternately executed with equal probability to improve the local escape ability of the algorithm.

[0024] For the above system, optionally, the feature analysis and pattern recognition results presented by the visualization display module include but are not limited to: time-domain signal, frequency-domain signal, time-frequency domain signal, distribution law of the first-level feature parameters, effective feature parameter set, vibration signal type, confusion matrix, and the time, position and number of broken wires.

[0025] It can be seen from the above technical solutions that compared with the prior art, the present invention provides a PCCP broken wire signal analysis system based on ISSA-HC:

[0026] (1) The advantages and applicable conditions of the filtering theory and the encapsulation theory are complementary. The present invention adopts the method of integrating the two theories, which can not only ensure the classification accuracy but also take into account the operation efficiency;

[0027] (2) By introducing the adaptive adjustment strategy into the SSA in the present invention, the ISSA has more discoverers in the early stage. As the number of iterations increases, the number of discoverers decreases adaptively, while the number of followers gradually increases, and the algorithm realizes the adaptive transformation from global search to local precise search;

[0028] (3) By introducing the Gaussian mutation perturbation and the Tent chaotic perturbation into the SSA in the present invention, the local escape ability of the ISSA is improved as a whole, the probability of the algorithm falling into the local optimum is greatly reduced, and the optimization ability of the algorithm is improved;

[0029] (4) The present invention converts the solution space of ISSA from continuous values to discrete values according to the conversion function, and then establishes an interface between HC and ISSA, using the classification accuracy rate of HC for vibration signals as the discrimination criterion for evaluating the quality of the feature parameter subset.

[0030] (5) The purpose of the present invention is to improve the accuracy rate of real-time monitoring of PCCP broken wire signals, and provide technical support for the health diagnosis and disaster warning of in-service PCCP projects. Brief Description of the Drawings

[0031] In order to more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the following will briefly introduce the drawings required for the description of the embodiments or the prior art. Obviously, the drawings in the following description are only the embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other drawings can also be obtained according to the provided drawings.

[0032] Figure 1 It is a structural framework diagram of a PCCP broken wire signal analysis system based on ISSA-HC disclosed by the present invention;

[0033] Figure 2 It is a structural framework diagram of a Φ-OTDR type signal monitoring module disclosed by the present invention;

[0034] Figure 3 It is the layout of PCCP in the prototype monitoring test disclosed by the present invention;

[0035] Figure 4 It is the cutting process of the prestressed steel wire disclosed by the present invention, where 4a is the cutting process and 4b is after cutting;

[0036] Figure 5 It is the time-domain signal disclosed by the present invention, where 5a is the time-domain signal of broken wire-1, 5b is the time-domain signal of broken wire-2, 5c is the time-domain signal of knocking, and 5d is the time-domain signal of noise;

[0037] Figure 6 It is the frequency-domain signal disclosed by the present invention, where 6a is the frequency-domain signal of broken wire-1, 6b is the frequency-domain signal of broken wire-2, and 6c is the frequency-domain signal of knocking;

[0038] Figure 7 It is the time-frequency domain signal of broken wire-1 disclosed by the present invention, where 7a is the high-frequency component d1, 7b is the high-frequency component d2, 7c is the high-frequency component d3, 7d is the high-frequency component d4, 7e is the high-frequency component d5, and 7f is the high-frequency component d6;

[0039] Figure 8 It is the distribution law of some time-domain and time-frequency domain characteristic parameters disclosed by the present invention, where 8a is the time-domain characteristic parameter T10 The distribution law, where 8b is the time-frequency domain characteristic parameter Z 3，6 The distribution law;

[0040] Figure 9 This is the ISSA-HC flow chart disclosed by the present invention;

[0041] Figure 10 This is the result of pattern recognition based on 4 groups of effective characteristic parameters. Among them, 10a is the result of pattern recognition based on the effective characteristic parameters mined by ISSA-HC, 10b is the result of pattern recognition based on the effective characteristic parameters mined by IPSO-HC, and 10c is the result of pattern recognition based on the effective characteristic parameters mined by SSA-HC. Detailed implementation manners

[0042] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

[0043] In this application, relational terms such as first and second are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any actual relationship or order between these entities or operations. The terms "comprising", "including" or any other variant thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements not only includes those elements, but also includes other elements not expressly listed, or further includes elements inherent to such process, method, article or device. Without further limitation, an element defined by the statement "including one..." does not exclude the existence of additional identical elements in the process, method, article or device including the element.

[0044] The present invention can be used in many general or special computing device environments or configurations. For example: personal computers, server computers, handheld devices or portable devices, tablet devices, multi-processor devices, distributed computing environments including any of the above devices or equipment, and so on.

[0045] Refer to Figure 1 As shown, the present invention discloses a PCCP broken wire signal analysis system based on ISSA-HC, including a Φ-OTDR type signal monitoring module, a feature calculation module, a feature optimization module, a vibration pattern recognition module, and a visualization display module connected in sequence;

[0046] Among them, the Φ-OTDR signal monitoring module is used to collect vibration signals during the operation of PCCP;

[0047] The feature calculation module is used to perform time-domain, frequency-domain, and time-frequency domain analyses on the collected vibration signals, and calculate the first-level feature parameters;

[0048] The feature optimization module is used to extract effective feature parameters from the first-level feature parameters;

[0049] The vibration mode recognition module identifies the type of vibration signal based on the set of effective feature parameters;

[0050] The visualization display module is used to intuitively present the feature analysis and pattern recognition results to the operator.

[0051] Furthermore, as Figure 2 shown, the Φ-OTDR signal monitoring module is based on a coherent detection design, including: a laser, a first beam splitter, an acousto-optic modulator, an erbium-doped fiber amplifier, a circulator, a second beam splitter, a photodetector, and a data acquisition card;

[0052] Among them, the output end of the laser is connected to the input end of the first beam splitter, the first output end of the first beam splitter is connected to the input end of the acousto-optic modulator, the output end of the acousto-optic modulator is connected to the input end of the erbium-doped fiber amplifier, the output end of the erbium-doped fiber amplifier is connected to the first port of the circulator, the sensing fiber is arranged closely along the inner wall of the PCCP, the output end of the sensing fiber is connected to the second port of the circulator, the third port of the circulator is connected to the first input end of the second beam splitter, the second output end of the first beam splitter is connected to the second input end of the second beam splitter, the output end of the second beam splitter is connected to the input end of the photodetector, and the output end of the photodetector is connected to the input end of the data acquisition card.

[0053] Specifically, in this embodiment, a PCCP wire break signal analysis system based on ISSA-HC disclosed by the present invention is verified by means of a prototype test. The PCCP adopted in this embodiment has a length of 5m, an inner diameter of 3.2m, a total core thickness of 245mm, an outer diameter of the steel barrel of 3.343m, a mortar protective layer thickness of 25mm, a steel cylinder thickness of 1.5mm, a prestressed steel wire diameter of 7mm, a prestressed steel wire area of 2350mm 2 / m, an elastic modulus of the prestressed steel wire of 205000MPa, and an ultimate tensile strength of the prestressed steel wire of 1620MPa. The test pipe section is as Figure 3 shown, consisting of 5 PCCPs, semi-buried underground, with the pipe orifices plugged with steel plates and then filled with water to simulate the water operation condition. The length of the foundation pit is about 29m, the width is about 4m, and the depth is about 2m.

[0054] During the water conveyance process of PCCP, vibration signals such as the fracture of prestressed steel wires, impact by floating objects, pipeline leakage, construction excavation, drilling, and water hammer can all cause distributed fiber optic vibration. Among them, the sound characteristics of the impact by floating objects and the fracture of prestressed steel wires are the most similar, and it is difficult to effectively distinguish them only by experience. In the experiment, the wire break signal under hydrogen embrittlement was simulated by manual cutting. As Figure 4 shown, a hammer was used to strike the inner wall of PCCP to simulate the collision of floating objects in water with PCCP. In addition, to simulate the situation of simultaneous fracture of prestressed steel wires, 2 working conditions were set during wire cutting, that is, only cutting 1 wire and cutting 2 wires simultaneously. The vibration signals were recorded as the wire break - 1 signal and the wire break - 2 signal respectively, where 4a is the cutting process and 4b is after cutting.

[0055] A total of 4 types of optical cables with specifications numbered A, B, C, and D were laid out in the experiment. The optical cables were fixed tightly to the inner wall of the pipeline using strong glue and epoxy resin to ensure that the sensing optical fiber vibrates in accordance with the pipe wall. The laying length of the optical cable is 10 km. Through knocking tests, it was found that optical cable C has the best sensitivity and signal - to - noise ratio, and the average positioning error is ±2 m. Therefore, optical cable C was selected for vibration signal monitoring in this experiment.

[0056] Furthermore, the feature calculation module is used to analyze the data in the data acquisition card and calculate the primary feature parameters, including a time - domain calculation unit, a frequency - domain calculation unit, and a time - frequency domain calculation unit connected in sequence.

[0057] Furthermore, the primary feature parameters include: time - domain feature parameters, frequency - domain feature parameters, and time - frequency domain feature parameters;

[0058] The time - domain calculation unit calculates the time - domain feature parameters by analyzing the time - domain signal. The frequency - domain calculation unit maps the time - domain signal to the frequency domain using the discrete Fourier transform, and then obtains the frequency - domain feature parameters. The time - frequency domain calculation unit performs multi - frequency decomposition on the time - domain signal based on the discrete wavelet transform and calculates the time - frequency domain feature parameters accordingly.

[0059] Specifically, a total of 60 wire break - 1 signals, 60 wire break - 2 signals, and 120 knocking signals were collected in the experiment. The time - domain signals are as Figure 5 shown, where 5a is the time - domain signal of wire break - 1, 5b is the time - domain signal of wire break - 2, 5c is the time - domain signal of knocking, and 5d is the time - domain signal of noise.

[0060] The time - domain signal was mapped to the frequency domain using the discrete Fourier transform, as Figure 6 shown, where 6a is the frequency - domain signal of wire break - 1, 6b is the frequency - domain signal of wire break - 2, and 6c is the frequency - domain signal of knocking. The time - domain signal was multi - frequency decomposed based on the discrete wavelet transform, and the high - frequency components d1, d2, d3, d4, d5, d6 were selected. Figure 7It is the time-frequency domain signal for broken wire - 1. Among them, 7a is the high-frequency component d1, 7b is the high-frequency component d2, 7c is the high-frequency component d3, 7d is the high-frequency component d4, 7e is the high-frequency component d5, and 7f is the high-frequency component d6.

[0061] Establish first-level characteristic parameters for the time domain, frequency domain, and time-frequency domain. The calculation formulas are shown in Tables 1 to 3, where: x h is the time series, w a is the frequency series, f a is the amplitude, a is the number of spectral lines. ε ρ,h is the ρ-th high-frequency wavelet component, h = 1, 2, …, H, and H is the signal length. A total of 122 first-level characteristic parameters are obtained through calculation. Among them, there are 17 time-domain parameters, 15 frequency-domain parameters, and 90 time-frequency domain parameters. Figure 8 This is the distribution law of some characteristic parameters disclosed by the present invention. Among them, 8a is the distribution law of the time-domain characteristic parameter T 10 and 8b is the distribution law of the time-frequency domain characteristic parameter Z 3，6 . It can be seen that when distinguishing broken wire - 1, broken wire - 2, and knocking signals, T 10 is more effective than Z 3，6 .

[0062] Table 1 Time-domain characteristic parameters

[0063]

[0064] Table 2 Frequency-domain characteristic parameters

[0065]

[0066] Table 3 Time-frequency domain characteristic parameters

[0067]

[0068] Furthermore, the feature optimization module includes a filtering optimization unit and a packaging optimization unit connected in sequence;

[0069] Among them, the output end of the filtering optimization unit is connected to the input end of the packaging optimization unit.

[0070] Furthermore, in the filtering optimization unit, the first-level characteristic parameters output by the feature calculation module are judged according to the Fisher criterion, and the second-level characteristic parameters that meet the judgment criterion are selected;

[0071] Specifically, the Fisher criterion uses the ratio of between-class variance and within-class variance as the judgment standard. The higher the F score, the smaller the distance between samples of the same class and the larger the distance between samples of different classes, and the better the classification performance. The calculation formula of the Fisher criterion is:

[0072]

[0073]

[0074]

[0075] Where: N represents the total number of samples; k = 1, 2, …, K, K represents the number of sample categories, and the corresponding categories are represented as C1, C2, …, C k ; N k represents the number of samples in the k-th category; x (d) represents the value of the sample x on the d-th feature; m (d) represents the mean value of all samples on the d-th feature; represents the mean value of the samples in the k-th category on the d-th feature; is the between-class variance; is the within-class variance.

[0076] Furthermore, in the encapsulated optimization unit, by improving SSA, an ISSA optimization strategy is established, and the classification accuracy of HC for vibration signals is used as the discrimination criterion for evaluating the quality of the feature parameter subset. On this basis, an ISSA-HC integration framework is constructed with the help of the encapsulated modeling theory, and the secondary feature parameters output by the filtering optimization unit are deeply mined according to ISSA-HC to establish an effective feature parameter set.

[0077] Furthermore, the improvement process of SSA is as follows: an adaptive adjustment strategy is introduced to enhance the global search ability of the algorithm in the early stage and the local search ability in the later stage, and the local escape ability of the algorithm is improved by alternately performing Gaussian mutation perturbation and Tent chaotic perturbation with equal probability.

[0078] Specifically, in the standard SSA, the ratio of discoverers to followers is set to 20% and 80%, resulting in insufficient global search ability of the algorithm in the early stage and weak local search ability in the later stage. To address this problem, the present invention introduces an adaptive adjustment strategy, which is specifically expressed as:

[0079]

[0080] PN = λ·N (5)

[0081] SN = (1 - λ)·N (6)

[0082] Where: N is the number of sparrows; PN is the number of discoverers; SN is the number of followers; r is the current iteration number; r maxis the maximum number of iterations; e is the scaling factor; λ is the ratio of discoverers to followers; α is the perturbation factor; Rand(0,1) is a random number within [0,1]. It can be seen from equations (4) to (6) that as the number of iterations increases, the number of discoverers gradually decreases, while the number of followers gradually increases, enabling ISSA to have a strong global search ability in the early stage and a strong local search ability in the later stage.

[0083] The expression for Gaussian mutation perturbation is:

[0084]

[0085] Where: is the -th element of the position vector of the v-th sparrow; Gaussian(X) is the position after Gaussian mutation perturbation; Norm(0,1) is a random number obeying the standard normal distribution. The expression for Tent chaotic mapping is:

[0086]

[0087] Where: Rand(0,1) is a random number within [0,1]; N is the number of particles in the chaotic sequence. After Bernoulli shift transformation, equation (8) can be expressed as:

[0088]

[0089] The expression for Tent chaotic perturbation is:

[0090] l′ = X min + (X max - X min ) · l (10)

[0091]

[0092]

[0093] Where: X max and X min are the upper and lower limits of the solution space respectively; Tent(X) is the position after Tent chaotic perturbation; δ is the direction coefficient, randomly taking 1 or -1; γ is the attenuation factor.

[0094] The solution space of ISSA is continuous. To solve the feature selection problem, the optimal solution needs to be converted from continuous values to discrete values. Based on this, an interface between ISSA and HC is established, and the conversion function is:

[0095]

[0096] Where: Rand(0,1) is a random number within [0,1].

[0097] The process of HC judging the quality of feature parameter subsets is as follows. Let n be the number of signal samples; S be the feature parameter set of n samples; S i is the characteristic parameter set of sample i; g is the number of characteristic parameters; s i,t For S i The t-th dimension feature parameter of S. i and S can be expressed as:

[0098] S i ={s i,1 s i,2 …s i,t …s i,g} (14)

[0099]

[0100] Introduce the label matrix m, that is:

[0101] m={m1 m2 … m t … m g} (16)

[0102] Where: m t The value of m is 0 or 1. t =1 means selecting the t-th dimension feature parameter, m t = 0 means giving up the t-th dimension feature parameter. i,t =m t ·s i,t , construct the matrix Y i :

[0103] Y i ={y i,1 y i,2 … y i,t … y i,g} (17)

[0104] The matrix Y can be established as:

[0105]

[0106] Take advantage of Y i and Y j The Euclidean distance θ i,j Characterize the similarity between samples i and j, that is:

[0107]

[0108] With the help of the class average distance R p,q Characterize the similarity between categories p and q, that is:

[0109]

[0110] Where: u and c are the number of samples in p and q respectively. It can be seen that when both u and c are 1, R p,q degenerates to θ i,j , so Equation (19) is a special case of Equation (20). HC adopts a bottom-up clustering strategy, and the expression of the evaluation function f is:

[0111]

[0112] Where: Acc is the recognition accuracy; Q is the number of selected feature parameters; G is the total number of feature parameters; η ∈ [0, 1] is the proportion of Acc in the whole formula. The calculation formula of Acc is:

[0113]

[0114] Where: TP is the true positive; FP is the false positive; FN is the false negative; TN is the true negative. Figure 9 is the flowchart of ISSA-HC disclosed in the present invention.

[0115] Furthermore, the feature analysis and pattern recognition results presented by the visualization display module include but are not limited to: time-domain signals, frequency-domain signals, time-frequency domain signals, distribution laws of primary feature parameters, effective feature parameter sets, vibration signal types, confusion matrices, and the time, position, and number of broken wires.

[0116] Specifically, 30 broken wire-1 signals, 30 broken wire-2 signals, and 60 knocking signals are randomly selected from all 240 groups of vibration signals. To test ISSA-HC, two encapsulated algorithms, SSA-HC and IPSO-HC, are established for comparative analysis. The parameters that need to be initialized for IPSO include: the number of particles q, the maximum number of iterations r max ; the parameters χ1, χ2, and σ of the inertia weight; the parameters c 1s , c 1e , c 2s and c 2e ; the position parameter γ; the velocity parameter η. The parameter settings are shown in Table 4. As can be seen from Table 5, ISSA-HC, IPSO-HC, and SSA-HC converge after 245, 358, and 453 iterations respectively, and 24, 31, and 55 effective feature parameters are mined respectively.

[0117] Table 4 Parameter settings of ISSA, SSA, and IPSO

[0118]

[0119] Table 5 Effective feature parameters mined by ISSA-HC, IPSO-HC, and SSA-HC

[0120]

[0121] Taking the remaining 30 groups of broken wire - 1 signals, 30 groups of broken wire - 2 signals and 60 groups of knocking signals as test samples, the identification performance of 4 groups of effective characteristic parameter sets for 3 types of vibration signals was tested. Figure 10 Figure 10 shows the pattern recognition results of 4 groups of effective characteristic parameters. Among them, 10a is the result of pattern recognition based on the effective characteristic parameters mined by ISSA - HC, 10b is the result of pattern recognition based on the effective characteristic parameters mined by IPSO - HC, and 10c is the result of pattern recognition based on the effective characteristic parameters mined by SSA - HC. In the confusion matrix, 1 represents the broken wire - 1 signal, 2 represents the broken wire - 2 signal, and 3 represents the knocking signal. As can be seen from Table 6, the classification accuracies of the effective characteristic parameters mined by ISSA - HC, IPSO - HC and SSA - HC for the broken wire - 1, broken wire - 2 and knocking signals are 95.83%, 84.17% and 80.83% respectively. Compared with SSA - HC and IPSO - HC, the effective characteristic parameters mined by ISSA - HC have better pattern recognition performance.

[0122] Table 6 Evaluation Indexes of Pattern Recognition Performance

[0123]

[0124] Each embodiment in this specification is described in a progressive manner. For the same or similar parts among the embodiments, reference can be made to each other. Each embodiment focuses on the differences from other embodiments. In particular, for the system or system embodiments, since they are basically similar to the method embodiments, the description is relatively simple. For the relevant parts, reference can be made to the partial description of the method embodiments. The systems and system embodiments described above are only illustrative. The units described as separate components may or may not be physically separated. The components shown as units may or may not be physical units, that is, they may be located in one place or distributed to multiple network units. Some or all of the modules can be selected according to actual needs to achieve the purpose of the solution of this embodiment. Those of ordinary skill in the art can understand and implement it without creative efforts.

[0125] The above description of the disclosed embodiments enables those skilled in the art to implement or use the present invention. Various modifications to these embodiments will be obvious to those skilled in the art. The general principles defined herein can be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention will not be limited to the embodiments shown herein, but will be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A PCCP broken wire signal analysis system based on ISSA-HC, characterized in that: It includes a Φ-OTDR type signal monitoring module, a feature calculation module, a feature optimization module, a vibration mode recognition module and a visual display module which are connected in sequence; Among them, the Φ-OTDR signal monitoring module is used to collect vibration signals during the operation of PCCP; The feature calculation module is used to analyze the collected vibration signal in the time domain, frequency domain and time-frequency domain, and calculate the primary feature parameters; The feature optimization module is used to mine effective feature parameters from the primary feature parameters; the feature optimization module includes a filtering optimization unit and a package optimization unit connected in sequence; the output end of the filtering optimization unit is connected to the input end of the package optimization unit; in the package optimization unit, the ISSA optimization strategy is established by improving the SSA, and the classification accuracy of the HC on the vibration signal is used as the criterion for judging the quality of the feature parameter subset. On this basis, the ISSA-HC integrated framework is constructed with the help of the package modeling theory, and the secondary feature parameters output by the filtering optimization unit are deeply mined according to the ISSA-HC to establish an effective feature parameter set; The vibration pattern recognition module identifies the type of vibration signal based on the effective feature parameter set; The visual display module is used to present the feature analysis and pattern recognition results to the operator intuitively.

2. The ISSA-HC-based PCCP broken wire signal analysis system according to claim 1, characterized in that: The Φ-OTDR signal monitoring module is designed based on coherent detection, including: laser, first beam splitter, acousto-optic modulator, erbium-doped fiber amplifier, circulator, second beam splitter, photoelectric detector and data acquisition card; Among them, the laser output end is connected to the first beam splitter input end, the first output end of the first beam splitter is connected to the acousto-optic modulator input end, the acousto-optic modulator output end is connected to the erbium-doped fiber amplifier input end, the erbium-doped fiber amplifier output end is connected to the first port of the circulator, the sensing fiber is arranged close to the inner wall of the PCCP, the sensing fiber output end is connected to the second port of the circulator, the third port of the circulator is connected to the first input end of the second beam splitter, the second output end of the first beam splitter is connected to the second input end of the second beam splitter, the output end of the second beam splitter is connected to the input end of the photodetector, and the output end of the photodetector is connected to the input end of the data acquisition card.

3. The ISSA-HC-based PCCP broken wire signal analysis system according to claim 1, characterized in that: The feature calculation module is used to analyze the data in the data acquisition card and calculate the primary feature parameters, including a time domain calculation unit, a frequency domain calculation unit and a time-frequency domain calculation unit connected in sequence.

4. The ISSA-HC-based PCCP broken wire signal analysis system according to claim 3, characterized in that: The first-level characteristic parameters include: time domain characteristic parameters, frequency domain characteristic parameters and time-frequency domain characteristic parameters; The time domain calculation unit calculates the time domain characteristic parameters by analyzing the time domain signal, and the frequency domain calculation unit uses discrete Fourier transform to map the time domain signal to the frequency domain to obtain the frequency domain characteristic parameters. The time-frequency domain calculation unit performs multi-frequency decomposition on the time domain signal according to discrete wavelet transform, and calculates the time-frequency domain characteristic parameters accordingly.

5. The ISSA-HC-based PCCP broken wire signal analysis system according to claim 1, characterized in that: Filtering optimization unit: The primary feature parameters output by the feature calculation module are judged according to the Fisher criterion, and the secondary feature parameters that meet the judgment criteria are selected.

6. The ISSA-HC-based PCCP broken wire signal analysis system according to claim 1, characterized in that: The improvement strategy of SSA is to introduce an adaptive adjustment strategy to enhance the algorithm's global search capability in the early stage and local search capability in the later stage, and to improve the algorithm's local escape ability by alternately executing Gaussian mutation perturbation and Tent chaos perturbation with equal probability.

7. The ISSA-HC-based PCCP broken wire signal analysis system according to claim 1, characterized in that: The feature analysis and pattern recognition results presented by the visualization module include: time domain signals, frequency domain signals, time-frequency domain signals, distribution law of primary feature parameters, effective feature parameter set, vibration signal type, confusion matrix, and the time, location and number of PCCP broken wires.

Images