Gate fault detection method and system

Abstract

The invention discloses a gate fault detection method and system, and relates to the technical field of fault diagnosis and health monitoring, and the method comprises the steps: collecting a gate vibration signal, extracting a main modal frequency, setting a sampling frequency and a passband range, designing a finite impulse response filter for filtering, and carrying out the optimization through a particle swarm algorithm, thereby obtaining an enhanced signal; extracting features in the enhanced signal for integration to obtain a feature vector, mapping the feature vector to a discrete state set by using a rule threshold matching method, and obtaining a gate state; the method comprises the following steps: acquiring historical data for sampling, constructing a state transition probability matrix, calculating a steady-state probability vector according to the state transition probability matrix, combining a gate state, generating a state sequence track, setting an influence factor and a weight by using an empirical rule, and calculating a fault occurrence probability for risk detection; according to the method, the gate fault detection precision, efficiency and reliability are remarkably improved.

Description

Technical Field

[0001] The present invention relates to the technical field of fault diagnosis and health monitoring, and particularly to a gate fault detection method and system. Background Art

[0002] With the continuous development of industrial automation, gates, as important control devices, play a crucial role in multiple fields such as water conservancy, transportation, and energy. In recent years, with the development of sensor technology, signal processing technology, and artificial intelligence technology, the fault detection method based on vibration signals has gradually become a mainstream solution. By collecting vibration signals through high-precision sensors and using frequency-domain analysis, filtering technology, and pattern recognition methods, the early faults of gates can be effectively detected.

[0003] However, the existing gate fault detection technologies have the following problems: Firstly, feature extraction is mostly based on single-domain analysis, ignoring time-frequency joint features and the non-stationary characteristics of signals. Secondly, state classification algorithms generally use static thresholds or shallow models, lacking the ability to adaptively fuse multi-dimensional features, and the threshold setting depends on empirical data, which is prone to misjudgment due to changes in working conditions. Thirdly, fault prediction models are mostly based on isolated state analysis, without considering the temporal correlation of historical data, making it difficult to construct a dynamic transition probability network. Summary of the Invention

[0004] In view of the above existing problems, the present invention is proposed.

[0005] Therefore, the present invention provides a gate fault detection method and system, which solve the problems in the prior art that existing feature extraction is mostly based on single-domain analysis, ignoring time-frequency joint features and the non-stationary characteristics of signals, state classification algorithms generally use static thresholds or shallow models, lacking the ability to adaptively fuse multi-dimensional features, and the threshold setting depends on empirical data, which is prone to misjudgment due to changes in working conditions, and fault prediction models are mostly based on isolated state analysis, without considering the temporal correlation of historical data, making it difficult to construct a dynamic transition probability network.

[0006] To solve the above technical problems, the present invention provides the following technical solutions: In the first aspect, the present invention provides a gate fault detection method, which includes, Collecting the gate vibration signal, extracting the main modal frequency, setting the sampling frequency and passband range, designing a finite impulse response filter for filtering, and then using the particle swarm algorithm for optimization to obtain an enhanced signal; Extracting and integrating the features in the enhanced signal to obtain a feature vector, using the rule threshold matching method to map the feature vector to a discrete state set, and obtaining the gate state; The gate state includes a normal state, a slight structural fatigue state, a structural offset abnormal state, and a critical structural damage state; Obtain historical data for sampling, construct a state transition probability matrix, calculate the steady-state probability vector based on the state transition probability matrix, combine the gate state, generate a state sequence trajectory, set influence factors and weights using empirical rules, calculate the probability of failure for risk detection; Store the data through a database.

[0007] As a preferred solution of the gate fault detection method described in the present invention, wherein: the steps of collecting the gate vibration signal, extracting the main modal frequency, setting the sampling frequency and passband range, designing a finite impulse response filter for filtering, and then using the particle swarm algorithm for optimization to obtain the enhanced signal include the following: Use a triaxial acceleration sensor to collect the original vibration signal of the gate; Perform frequency domain analysis on the collected vibration signal using the fast Fourier transform, and calculate the amplitude of the frequency domain signal using the absolute value; Use the frequency main peak extraction algorithm to select the frequency corresponding to the maximum amplitude as the main modal frequency ; Set the sampling frequency using the Nyquist theorem and the sampling duration ; Set the passband range using the main modal frequency, and extract the passband width using subtraction operation; Design a finite impulse response filter using the window function method, and construct the finite impulse response filter coefficients using the Hanning window function; Calculate the order of the finite impulse response filter using the order empirical formula; Perform a convolution operation on the vibration signal using the constructed finite impulse response filter to obtain a filtered signal, obtain the mean and standard deviation of the filtered signal, and calculate the kurtosis value; Use the delay embedding technique to convert the filtered signal into embedding vectors at different time points, and use the Chebyshev distance to traverse all embedding vectors, calculate the distance between the current embedding vector and all embedding vectors, sort all the distances in ascending order, and select the maximum distance to define the similarity; Calculate the entropy value of the filtered signal using the fuzzy entropy algorithm combined with the similarity; Set the initial weight coefficient using the empirical rule, and calculate the weight coefficients of the kurtosis value and the entropy value respectively using the proportional formula; Integrate the kurtosis value, the entropy value and the weight coefficient as particles, and randomly generate the initial position and population for initialization; Construct an objective function using the kurtosis value, the entropy value and the weight coefficient; Calculate the objective function value, and use the objective function value as the fitness value of the particle. Select the maximum fitness value for position update. During the iteration process, when the number of iterations reaches the maximum number, output the optimal weight coefficient; Take the optimal weight coefficient as the finite impulse response filter parameter, and perform deconvolution operation using the blind deconvolution technique to obtain the enhanced signal and the amplitude at each time point.

[0008] As a preferred solution of the gate fault detection method described in the present invention, wherein: extracting and integrating the features in the enhanced signal to obtain a feature vector, using the rule threshold matching method to map the feature vector to a discrete state set, and obtaining the gate state includes the following steps: Calculate the energy index of the enhanced signal using the weighted sum of squares formula; Use the Welch method to divide the enhanced signal into overlapping segments, construct the periodogram of each overlapping segment, and average all the overlapping segments to obtain the power spectral density; Calculate the bandwidth power ratio of the enhanced signal using the frequency domain analysis method combined with the power spectral density; Further use the fuzzy entropy algorithm to calculate the entropy value of the enhanced signal; Use the feature splicing technique to splice the energy index, bandwidth power ratio and entropy value to obtain a feature vector for normalization operation; Use the rule threshold matching method to map the feature vector to the discrete state set, and perform state conversion during the mapping process to obtain the current state of the gate; The discrete state set includes the normal state , the slight structural fatigue state , the structural offset abnormal state and the critical structural damage state ; Use the Delphi method to set the priority of the states to obtain ; During the state conversion process, when the state satisfies multiple states at the same time, then determine the maximum priority as the preferred state according to the priority.

[0009] As a preferred solution of the gate fault detection method described in the present invention, wherein: sampling the obtained historical data, constructing a state transition probability matrix, calculating the steady-state probability vector according to the state transition probability matrix, combining the gate state, generating a state sequence trajectory, and using the empirical rule to set the influence factor and weight to calculate the fault occurrence probability includes the following steps: Collect the historical fault vibration data of the gate; Take the current state of the gate as the starting point, use the Metropolis random walk algorithm to sample the state sequence from the historical operation data, and construct a state change sequence; After calculating the transition probabilities between the current state of the gate and all states in the state change sequence using the Markov chain model, a state transition probability matrix is constructed. The power iteration method is used to iteratively solve each row of the state transition probability vectors in the state transition probability matrix to obtain the steady-state probability vectors. Randomly sample a candidate state from the transition probability matrix , and calculate the acceptance probability. Use a pseudo-random number generator to generate a random number, and compare the acceptance probability with the random number. When the acceptance probability is greater than or equal to the random number, the candidate state is taken as the new state; otherwise, the current state is retained. Repeat sampling to generate a state sequence trajectory, and use the damage state corresponding to each state in the state sequence trajectory as a fault cause node. Use empirical rules to set the influence factor and weight respectively. Use the weighted summation formula to calculate the influence score. Use the Sigmoid function to map the influence score to obtain the influence probability. Use DeMorgan's theorem combined with the probability product rule to calculate the fault occurrence probability.

[0010] As a preferred solution of the gate fault detection method of the present invention, wherein: the risk detection further refers to using empirical rules to set an evaluation threshold, comparing the fault occurrence probability with the evaluation threshold. When the fault occurrence probability is greater than the evaluation threshold, it indicates that the current situation is at high risk, triggering an alarm and reminding the staff to take emergency repair measures; otherwise, continue the detection and generate a gate health report.

[0011] As a preferred solution of the gate fault detection method of the present invention, wherein: collect the vibration data and first perform preprocessing operations. The preprocessing operations include interpolation, smoothing, and normalization.

[0012] As a preferred solution of the gate fault detection method of the present invention, wherein: the storage of data through the database means storing the detection results, feature vectors, and state sequence trajectories into the database respectively, and adding time stamps to each data using the database and then classifying them.

[0013] In a second aspect, the present invention provides a gate fault detection system, including An acquisition and optimization module, used to acquire the gate vibration signal, extract the main modal frequency, set the sampling frequency and passband range, design a finite impulse response filter for filtering, and then use the particle swarm algorithm for optimization to obtain an enhanced signal. A detection module, which is used to extract and integrate features in the enhanced signal to obtain a feature vector, map the feature vector to a discrete state set using a rule threshold matching method, and obtain the gate state; An adoption and generation module, which is used to obtain historical data for sampling, construct a state transition probability matrix, calculate a steady-state probability vector according to the state transition probability matrix, combine the gate state, generate a state sequence trajectory, set influence factors and weights using empirical rules, and calculate the probability of a fault occurring for risk detection; A storage module, which is used to store data through a database.

[0014] In a third aspect, the present invention provides a computer device, including a memory and a processor, where the memory stores a computer program, and: when the computer program is executed by the processor, any step of the gate fault detection method described in the first aspect of the present invention is implemented.

[0015] In a fourth aspect, the present invention provides a computer-readable storage medium, on which a computer program is stored, and: when the computer program is executed by the processor, any step of the gate fault detection method described in the first aspect of the present invention is implemented.

[0016] The beneficial effects of the present invention are as follows: By combining time-domain kurtosis analysis and fuzzy entropy calculation, and further integrating multi-dimensional features such as energy index and bandwidth power ratio, the present invention solves the limitation of traditional methods relying on single-domain analysis. And by splicing multiple features such as kurtosis and entropy value into a feature vector, and combining the fuzzy entropy algorithm and the rule threshold matching method for state mapping, the accuracy and robustness of state classification are greatly improved. Secondly, a time-series analysis method based on historical data is introduced, and the Metropolis random walk algorithm and the Markov chain model are used to construct a dynamic transition probability network, which can effectively capture the time-series relationship of historical data, and then realize the dynamic prediction of the occurrence of faults. Therefore, by comprehensively considering time-frequency joint features, optimizing states and classification, and introducing time-series correlation analysis, the present invention significantly improves the accuracy, efficiency and reliability of gate fault detection. Description of the Drawings

[0017] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following will briefly introduce the drawings required for the description of the embodiments. Obviously, the following drawings are only some embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other drawings can also be obtained based on these drawings.

[0018] Figure 1 It is a flowchart of the gate fault detection method in Embodiment 1.

[0019] Figure 2It is the structural diagram of the gate fault detection system in Embodiment 1.

[0020] Figure 3 It is the flowchart of signal processing in Embodiment 1.

[0021] Figure 4 It is the flowchart of gate status judgment in Embodiment 1. Detailed implementation manners

[0022] To make the above objects, features, and advantages of the present invention more obvious and understandable, the following will describe the detailed implementation manners of the present invention with reference to the accompanying drawings of the specification.

[0023] In the following description, many specific details are set forth to facilitate a full understanding of the present invention. However, the present invention can also be implemented in other ways different from those described herein. Those skilled in the art can make similar extensions without departing from the connotation of the present invention. Therefore, the present invention is not limited by the specific embodiments disclosed below.

[0024] Secondly, the so-called "one embodiment" or "embodiment" herein refers to a specific feature, structure, or characteristic that can be included in at least one implementation manner of the present invention. The phrase "in one embodiment" appearing in different places in this specification does not necessarily refer to the same embodiment, nor is it a separate or alternative embodiment that is mutually exclusive with other embodiments.

[0025] Embodiment 1, referring to Figures 1 to 4 , which is the first embodiment of the present invention. This embodiment provides a gate fault detection method, including the following steps: S1. Collect the vibration signal of the gate, extract the main modal frequency, set the sampling frequency and passband range, design a finite impulse response filter for filtering, and then use the particle swarm algorithm for optimization to obtain an enhanced signal; Specifically, collecting the vibration signal of the gate, extracting the main modal frequency, setting the sampling frequency and passband range, designing a finite impulse response filter for filtering, and then using the particle swarm algorithm for optimization to obtain an enhanced signal includes the following steps: Use a triaxial acceleration sensor to collect the original vibration signal of the gate; Use the fast Fourier transform to perform frequency domain analysis on the collected vibration signal, and use the absolute value to calculate the amplitude of the frequency domain signal; Use the frequency main peak extraction algorithm to select the frequency corresponding to the maximum amplitude as the main modal frequency ; Use the Nyquist theorem to set the sampling frequency and the sampling duration ; Use the main modal frequency to set the passband range, and use subtraction operation to extract the passband width; The lower frequency limit of the passband range is: , and the upper frequency limit of the passband range is: ; Design a finite impulse response filter using the window function method and construct the finite impulse response filter coefficients using the Hann window function; Calculate the order of the finite impulse response filter using the order empirical formula, and the formula is:

[0026] wherein, represents the order of the finite impulse response filter, represents the sampling frequency, represents the passband width, represents rounding up; Perform a convolution operation on the vibration signal using the constructed finite impulse response filter to obtain a filtered signal, obtain the mean and standard deviation of the filtered signal, and calculate the kurtosis value. The formula is:

[0027] wherein, represents the kurtosis value of the filtered signal , represents the standard deviation of the filtered signal , represents the mean of the filtered signal , represents the total length of the filtered signal, represents the filtered signal the th amplitude; Use the delay embedding technique to convert the filtered signal into embedding vectors at different time points, and use the Chebyshev distance to traverse all the embedding vectors, calculate the distances between the current embedding vector and all the embedding vectors, and after sorting all the distances in ascending order, select the maximum distance to define the similarity; Use the fuzzy entropy algorithm combined with the similarity to calculate the entropy value of the filtered signal; Set the initial weight coefficient using the empirical rule, and calculate the weight coefficients of the kurtosis value and the entropy value respectively using the proportional formula. The formula is:

[0028]

[0029] wherein, represents the weight coefficient of the kurtosis value, represents the weight coefficient of the entropy value, and respectively represent the initial weight coefficients of kurtosis value and entropy value; Integrate the kurtosis value, entropy value and weight coefficient as particles, and randomly generate the initial positions and populations for initialization; Use the kurtosis value, entropy value and weight coefficient to construct the objective function, and the formula is:

[0030] In the formula, represents the objective function value of the filtered signal , represents the entropy value of the filtered signal , represents the natural logarithm; Calculate the objective function value, and use the objective function value as the fitness value of the particle. Select the maximum fitness value for position update. During the iteration process, when the number of iterations reaches the maximum number, output the optimal weight coefficient; Use the optimal weight coefficient as the finite impulse response filter parameter, and perform deconvolution operation using the blind deconvolution technique to obtain the enhanced signal and the amplitude at each time point.

[0031] By simultaneously collecting vibration signals in three directions through a triaxial acceleration sensor, the vibration characteristics of the gate can be comprehensively captured. Using FFT for frequency-domain analysis of the vibration signals and finding the main modal frequency through the main peak extraction algorithm can significantly improve the understanding of the frequency characteristics of the signals. By accurately obtaining the main modal frequency, the present invention can be optimized for specific frequencies, improving the sensitivity of the present invention to potential failure modes. Using the Nyquist theorem to set the sampling frequency can ensure that sufficient information is collected, avoiding signal distortion or aliasing caused by insufficient sampling. This ensures that the filtered signal is more real and reliable. And by designing an FIR filter using the window function method and optimizing the filter parameters in combination with the particle swarm optimization algorithm, the performance of the filter can be effectively improved, removing the noise components in the signal and retaining the useful vibration characteristics. The present invention not only improves the quality of the signal but also improves the accuracy of signal analysis, especially having significant advantages in an environment with large noise interference. Secondly, by calculating the kurtosis and entropy values of the filtered signal and combining and optimizing them with the weight coefficients, the filtered signal can be made to better conform to the target optimization criteria. And by optimizing the weight coefficients using the particle swarm algorithm, the present invention can adaptively adjust the parameters to enhance the signal in an optimal manner, further improving the adaptive ability of the system. And using the blind deconvolution technique to perform deconvolution operations on the filtered signal can extract a more accurate and clear enhanced signal from the processed signal. This process helps to restore the original characteristics of the vibration signal by removing system noise and distortion, providing more accurate data support for subsequent fault diagnosis and prediction. Therefore, the present invention not only improves the accuracy of signal collection and analysis but also greatly enhances the efficiency and effect in the filtering and enhancement processes. At the same time, combining the fuzzy entropy algorithm with kurtosis calculation optimizes the signal feature extraction, enabling the present invention to more accurately capture the abnormal features in the vibration signal.

[0032] S2. Extract the features in the enhanced signal for integration to obtain a feature vector, use the rule threshold matching method to map the feature vector to a discrete state set, and obtain the gate state; the gate state includes a normal state, a slight structural fatigue state, a structural offset abnormal state, and a critical structural damage state; Specifically, extracting the features in the enhanced signal for integration to obtain a feature vector, using the rule threshold matching method to map the feature vector to a discrete state set, and obtaining the gate state includes the following steps: Use the weighted sum of squares formula to calculate the energy index of the enhanced signal, and the formula is:

[0033] In the formula, represents the energy index, represents the amplitude of the enhanced signal at time point of, Indicates the total number of time points; Use the Welch method to divide the enhanced signal into overlapping segments, construct the periodogram of each overlapping segment, and average all overlapping segments to obtain the power spectral density; Use the frequency-domain analysis method in combination with the power spectral density to calculate the bandwidth power ratio of the enhanced signal. The formula is:

[0034] In the formula, Indicates the bandwidth power ratio, Indicates the power spectral density, Indicates the main modal frequency; Further use the fuzzy entropy algorithm to calculate the entropy value of the enhanced signal; Use the feature splicing technology to splice the energy index, bandwidth power ratio, and entropy value to obtain a feature vector for normalization; Use the rule threshold matching method to map the feature vector to the discrete state set and perform state conversion during the mapping process to obtain the current state of the gate; The discrete state set includes the normal state , the slight structural fatigue state , the structural offset abnormal state and the critical structural damage state ; Use the Delphi method to set the priority of the states to obtain ; During the state conversion process, when the state satisfies multiple states at the same time, the maximum priority is determined as the priority state according to the priority; The judgment condition formula for state conversion during the mapping process is:

[0035] In the formula, and respectively represent the judgment thresholds of the slight structural fatigue state, represents the judgment threshold of the structural offset abnormal state, and represent the judgment thresholds of the critical structural damage state, represents the enhanced signal entropy value.

[0036] The energy index of the enhanced signal is calculated through the weighted sum of squares formula, enabling the present invention to quantify the overall energy of the signal, thereby providing effective support for subsequent state classification. The enhanced signal is divided into overlapping segments using the Welch method, and the power spectral density of each segment is calculated. Through the calculation of the power spectral density, the frequency characteristics of the signal can be revealed, and valuable frequency information can be extracted therefrom, which is crucial for evaluating the vibration characteristics of the gate, especially in the early fault state identification of slight fatigue, abnormal offset, and critical damage states. As an important indicator for signal frequency domain analysis, the bandwidth power ratio can reflect the frequency concentration and bandwidth characteristics of the signal through its calculation. In the health monitoring of the gate, changes in the bandwidth power ratio can indicate potential structural problems or damage areas. Through this feature, the sensitivity of fault diagnosis can be improved, especially providing a reliable basis for identifying different types of fault states. By calculating the entropy value of the signal using the fuzzy entropy algorithm, it helps to reveal the complexity and uncertainty of the signal. The higher the entropy value, the more uncertain the signal changes, which is usually related to potential structural abnormalities or fault states. Combining the calculation of the entropy value can identify the abnormal state of the gate earlier, thereby enhancing the ability of the present invention to perceive early faults. Secondly, the energy index, bandwidth power ratio, and entropy value are concatenated to obtain a comprehensive feature vector, and then normalized processing is performed, which can effectively remove the scale differences between data, enabling different features to have the same comparison weight. This operation not only improves the stability of the model but also optimizes the accuracy of state classification, ensuring that various features can equally affect the final state determination. And the rule threshold matching method maps the feature vector to a discrete state set through pre-set rules and performs precise classification during the state transition process. Through this method, the present invention can accurately divide between different states and determine the working state of the current gate according to the set rules, such as normal state, slight structural fatigue state, structural offset abnormal state, and critical structural damage state. The introduction of this step greatly improves the reliability of state determination, especially when multiple fault states coexist, it can efficiently distinguish and determine. At the same time, the Delphi method scientifically sets the priorities of each state through expert opinions to ensure that in the case of multiple states, the fault state that poses the greatest threat to the gate can be identified first.

[0037] S3. Obtain historical data for sampling, construct a state transition probability matrix, calculate the steady-state probability vector according to the state transition probability matrix, combine with the gate state to generate a state sequence trajectory, set the influence factor and weight using empirical rules, and calculate the probability of failure for risk detection; Specifically, obtaining historical data for sampling, constructing a state transition probability matrix, calculating the steady-state probability vector according to the state transition probability matrix, combining with the gate state to generate a state sequence trajectory, and setting the influence factor and weight using empirical rules, calculating the probability of failure includes the following steps: Collect the historical fault vibration data of the gate; Taking the current state of the gate as the starting point, use the Metropolis random walk algorithm to sample the state sequence from the historical operation data and construct a state change sequence; After calculating the transition probabilities between the current state of the gate and all states in the state change sequence using the Markov chain model, construct a state transition probability matrix; Use the power iteration method to iteratively solve each row of the state transition probability vectors in the state transition probability matrix to obtain the steady-state probability vector; Randomly sample a candidate state from the transition probability matrix , and calculate the acceptance probability, the formula is:

[0038] In the formula, represents the acceptance probability, represents the current state of the gate, and respectively represent the steady-state probability vectors of the gate states and , represents the minimum value operation; Use a pseudo-random number generator to generate a random number and compare the acceptance probability with the random number. When the acceptance probability is greater than or equal to the random number, then use the candidate state as the new state, otherwise retain the current state; Repeat sampling to generate a state sequence trajectory, and use the damage state corresponding to each state in the state sequence trajectory as a fault cause node; Use empirical rules to set the influence factor and weight respectively; Use the weighted summation formula to calculate the influence score, the formula is:

[0039] In the formula, represents the influence score of the th fault cause node , represents the weight of the energy index, represents the weight of the power spectral density, represents the weight of the enhanced signal entropy value; Use the Sigmoid function to map the influence score to obtain the influence probability, the formula is:

[0040] In the formula, represents the th fault cause node after mapping The influence probability represents the base of the natural logarithm; Using DeMorgan's theorem in combination with the probability multiplication rule, calculate the probability of failure occurrence. The formula is:

[0041] In the formula, represents the probability of failure occurrence under the given failure cause node and the suppression factor condition. When represents the failure state, it indicates that the failure occurs; when , it indicates that the failure does not occur. represents the th suppression factor 's adjustment weight. represents the influence factor. represents the th failure cause node 's adjustment weight. The suppression factor and the adjustment weight as well as the adjustment weight can be set based on expert experience and experimental data.

[0042] By collecting the historical fault vibration data of the gate and performing state sequence sampling, a state transition model of the system can be constructed based on the historical fault modes to predict future state changes. This not only improves the accuracy of fault prediction but also provides data support for the subsequent construction of the state transition probability matrix. The sampled state sequence provides a solid foundation for subsequent fault prediction, enabling the present invention to detect potential fault points in advance and avoid system fault failures. The Metropolis algorithm, through the idea of Markov chain, effectively avoids excessive unnecessary calculations and can generate a state sequence that conforms to the actual state change law according to historical data, providing an efficient sampling method. Moreover, by accepting or rejecting the sampling results, this algorithm helps to construct a state change trajectory that conforms to the actual operation law of the system, thus improving the accuracy and reliability of model prediction. The construction of the state transition probability matrix is a crucial step in the entire fault prediction process. By analyzing the transition probabilities between different states of the system, this step helps to identify which state transitions have a higher probability, so that the future state of the system can be judged more accurately during prediction. Secondly, by modeling the state transition through the Markov chain model, the probability of transitioning from the current state to the fault state can be effectively predicted, thus providing decision-making support for fault prediction and prevention. The calculation of the steady-state probability vector can reveal the stable state distribution under long-term operation, which provides a clear criterion for fault prediction. For example, if the steady-state probability of a certain state is high, it indicates that this state may be the most frequently occurring state during stable operation. If this state is close to the fault state, it indicates that there may be a high fault risk. By efficiently calculating the steady-state probability vector through the power iteration method, the long-term stable state of the system can be quickly and accurately identified, and then the fault tendency of the system can be predicted. By combining the influencing factors and weights to perform weighted summation on the fault causes, the present invention can comprehensively consider the weights of various influencing factors and provide more accurate fault prediction results. Moreover, using the Sigmoid function to map the influence score to the influence probability can more accurately estimate the influence of each fault cause node on the occurrence of the fault, thus providing a more scientific fault prevention plan for decision-makers. And by combining DeMorgan's theorem and the probability product rule to calculate the fault occurrence probability, the present invention can comprehensively consider the combined influence of multiple factors (such as fault causes, suppression factors) on the occurrence of the fault, which improves the accuracy of fault prediction and enhances the robustness of the system. Through this probability calculation framework, the present invention can flexibly adjust the weights of various factors to adapt to different fault modes in different situations, providing guidance for maintenance and optimization.

[0043] Further, risk detection means further setting an evaluation threshold using empirical rules, comparing the fault occurrence probability with the evaluation threshold. When the fault occurrence probability is greater than the evaluation threshold, it indicates that the current situation is at high risk, triggering an alarm and reminding the staff to take emergency repair measures. Otherwise, continue the detection and generate a gate health report.

[0044] By introducing a risk detection mechanism, the present invention can predict in advance the possible failures of the device and issue early warnings in a timely manner. Such early warnings help to avoid sudden failures of the device, ensure the long-term stable operation of the device. When the probability of failure is higher than the evaluation threshold, an alarm will be automatically triggered and emergency repair measures will be recommended, reducing human intervention and delays, and improving the response speed and efficiency of fault handling. In addition, through the generation of health reports, maintenance personnel can have a more comprehensive understanding of the operating conditions of the device, thereby making more reasonable maintenance decisions.

[0045] S4. Store the data through a database; Specifically, storing the data through a database means storing the detection results, feature vectors, and state sequence trajectories into the database respectively, and adding timestamps to each data using the database and then classifying them.

[0046] Storing the detection results, feature vectors, and state sequence trajectories through the database enables a large amount of data to be effectively managed in a structured manner. By introducing timestamps, each data item can be accurately calibrated in time, making the data storage process both efficient and orderly.

[0047] This embodiment also provides a gate fault detection system, including: An acquisition and optimization module, which is used to acquire gate vibration signals, extract the main modal frequencies, set the sampling frequency and passband range, design a finite impulse response filter for filtering, and then use the particle swarm algorithm for optimization to obtain enhanced signals; A detection module, which is used to extract and integrate the features in the enhanced signals to obtain feature vectors, map the feature vectors to a discrete state set using the rule threshold matching method, and obtain the gate state; An adoption and generation module, which is used to acquire historical data for sampling, construct a state transition probability matrix, calculate the steady-state probability vector according to the state transition probability matrix, combine the gate state, generate a state sequence trajectory, set the influence factor and weight using the empirical rule, and calculate the probability of failure for risk detection; A storage module, which is used to store the data through a database.

[0048] This embodiment also provides a computer device applicable to the gate fault detection method, including: a memory and a processor; the memory is used to store computer-executable instructions, and the processor is used to execute the computer-executable instructions to implement the gate fault detection method proposed in the above embodiment.

[0049] The computer device may be a terminal, which includes a processor, a memory, a communication interface, a display screen, and an input device connected via a system bus. Among them, the processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and computer programs. The internal memory provides an environment for the operation of the operating system and computer programs in the non-volatile storage medium. The communication interface of the computer device is used to communicate with an external terminal in a wired or wireless manner. The wireless manner can be achieved through WIFI, a carrier network, NFC (Near Field Communication), or other technologies. The display screen of the computer device can be a liquid crystal display screen or an electronic ink display screen. The input device of the computer device can be a touch layer covering the display screen, or a button, a trackball, or a touchpad provided on the housing of the computer device, or an external keyboard, touchpad, or mouse, etc.

[0050] This embodiment also provides a storage medium on which a computer program is stored. When the program is executed by a processor, it implements the method for detecting gate faults as proposed in the above embodiment; the storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as Static Random Access Memory (SRAM), Electrically Erasable Programmable Read-Only Memory (EEPROM), Erasable Programmable Read Only Memory (EPROM), Programmable Read-Only Memory (PROM), Read-Only Memory (ROM), magnetic memory, flash memory, a magnetic disk, or an optical disc.

[0051] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit them. Although the present invention has been described in detail with reference to the preferred embodiments, those of ordinary skill in the art should understand that the technical solutions of the present invention can be modified or equivalently replaced without departing from the spirit and scope of the technical solutions of the present invention, and they should all be covered by the scope of the claims of the present invention.

Claims

1. A gate fault detection method, characterized in that: include, Collect gate vibration signals, extract main modal frequencies, set sampling frequency and passband range, design finite impulse response filters for filtering, and use particle swarm optimization to obtain enhanced signals. Extract the features from the enhanced signal and integrate them to obtain the feature vector. Use the rule threshold matching method to map the feature vector to a discrete state set and obtain the gate state. The gate status includes a normal state, a slight structural fatigue state, an abnormal structural deviation state, and a critical structural damage state; Obtain historical data for sampling and construct a state transition probability matrix. Calculate the steady-state probability vector based on the state transition probability matrix. Combine the gate state to generate the state sequence trajectory. Use empirical rules to set the influencing factors and weights, and calculate the probability of failure for risk detection. The data is stored in a database.

2. The gate fault detection method according to claim 1, characterized in that: The method of collecting the gate vibration signal, extracting the main modal frequency, setting the sampling frequency and the passband range, designing a finite impulse response filter for filtering, and optimizing using a particle swarm algorithm to obtain an enhanced signal includes the following steps: Use a three-axis acceleration sensor to collect the original vibration signal of the gate; The collected vibration signal is analyzed in the frequency domain using fast Fourier transform, and the amplitude of the frequency domain signal is calculated using the absolute value; Use the frequency main peak extraction algorithm to select the frequency corresponding to the maximum amplitude as the main modal frequency ; Using Nyquist theorem to set sampling frequency and sampling duration ; Use the dominant mode frequency to set the passband range and use a subtraction operation to extract the passband width; Design the FIR filter using the window function method and construct the FIR filter coefficients using the Hanning window function; Use the empirical formula to calculate the order of the finite impulse response filter; Use the constructed finite impulse response filter to perform convolution operation on the vibration signal to obtain a filtered signal, obtain the mean and standard deviation of the filtered signal, and calculate the kurtosis value; Use delayed embedding technology to convert the filtered signal into embedding vectors at different time points, and use Chebyshev distance to traverse all embedding vectors, calculate the distance between the current embedding vector and all embedding vectors, sort all distances in ascending order, and select the largest distance to define as similarity; The entropy value of the filtered signal is calculated using the fuzzy entropy algorithm combined with similarity; Use the empirical rule to set the initial weight coefficient, and use the proportional formula to calculate the weight coefficients of the kurtosis value and entropy value respectively; Integrate kurtosis value, entropy value and weight coefficient as particles, and randomly generate initial positions and populations for initialization; Use kurtosis value, entropy value and weight coefficient to construct the objective function; Calculate the objective function value and use it as the fitness value of the particle. Select the maximum fitness value to update the position. During the iteration process, when the number of iterations reaches the maximum, output the optimal weight coefficient. The optimal weight coefficients are used as finite impulse response filter parameters, and the deconvolution operation is performed using blind deconvolution technology to obtain the enhanced signal and the amplitude at each time point.

3. The gate fault detection method according to claim 2, characterized in that: The extraction and integration of features in the enhanced signal to obtain a feature vector, the use of a rule threshold matching method to map the feature vector to a discrete state set, and the acquisition of the gate state includes the following steps: The energy index of the enhanced signal is calculated using the weighted sum of squares formula; The enhanced signal is divided into overlapping segments using the Welch method, the periodogram of each overlapping segment is constructed, and all overlapping segments are averaged to obtain the power spectral density; The bandwidth power ratio of the enhanced signal is calculated using frequency domain analysis combined with power spectral density; The fuzzy entropy algorithm is further used to calculate the entropy value of the enhanced signal; The energy index, bandwidth power ratio and entropy value are spliced ​​using feature splicing technology to obtain feature vectors for normalization operation; The feature vector is mapped to a discrete state set using the rule threshold matching method, and the state conversion is performed during the mapping process to obtain the current state of the gate; The discrete state set includes a normal state , Slight structural fatigue state , Structural deviation abnormal state and critical structural damage state ; Use the Delphi method to prioritize the states and get ; During the state conversion process, when a state satisfies multiple states at the same time, the highest priority is determined as the priority state according to the priority.

4. The gate fault detection method according to claim 3, characterized in that: The acquisition of historical data for sampling and the construction of a state transition probability matrix are performed. The steady-state probability vector is calculated according to the state transition probability matrix. The state sequence trajectory is generated by combining the gate state. The influencing factors and weights are set using empirical rules. The calculation of the probability of a fault occurrence includes the following steps: Collect historical fault vibration data of gates; Taking the current state of the gate as the starting point, the Metropolis random walk algorithm is used to sample the state sequence from the historical operation data and construct the state change sequence; After using the Markov chain model to calculate the transition probabilities between the current state of the gate and all states in the state change sequence, a state transition probability matrix is ​​constructed; Use the power iteration method to iteratively solve each row of the state transition probability vector in the state transition probability matrix to obtain the steady-state probability vector; Randomly sample a candidate state from the transition probability matrix , calculate the acceptance probability; Use a pseudo-number random generator to generate a random number, and compare the acceptance probability with the random number. When the acceptance probability is greater than or equal to the random number, the candidate state as the new state, otherwise keep the current state; Sampling is repeated to generate a state sequence trajectory, and the damage state corresponding to each state in the state sequence trajectory is used as the fault cause node; Use rules of thumb to set impact factors and weights respectively; The impact score is calculated using a weighted sum formula; Use the Sigmoid function to map the impact score to obtain the impact probability; Use DeMorgan's theorem combined with the probability product rule to calculate the probability of failure.

5. The gate fault detection method according to claim 4, characterized in that: The risk detection refers to further using empirical rules to set an evaluation threshold, comparing the probability of a failure with the evaluation threshold. When the probability of a failure is greater than the evaluation threshold, it indicates that the current risk is high, triggering an alarm and reminding staff to take emergency repair measures. Otherwise, continue detection and generate a gate health report.

6. The gate fault detection method according to claim 1, characterized in that: Collect the vibration data and perform a pre-processing operation first; The preprocessing operations include interpolation, smoothing and normalization.

7. The gate fault detection method according to claim 1, characterized in that: The storing of data through a database refers to storing the detection results, feature vectors and state sequence trajectories in the database respectively, and using the database to add a timestamp to each data for classification.

8. A gate fault detection system, based on the gate fault detection method according to claim 1, characterized in that: include, The acquisition and optimization module is used to acquire the gate vibration signal, extract the main modal frequency, set the sampling frequency and passband range, design a finite impulse response filter for filtering, and use the particle swarm algorithm for optimization to obtain an enhanced signal; The detection module is used to extract the features in the enhanced signal and integrate them to obtain the feature vector, map the feature vector to a discrete state set using the rule threshold matching method, and obtain the gate state; The generation module is used to obtain historical data for sampling and construct a state transition probability matrix. The steady-state probability vector is calculated based on the state transition probability matrix. The state sequence trajectory is generated by combining the gate state. The influencing factors and weights are set using empirical rules, and the probability of fault occurrence is calculated for risk detection. The storage module is used to store data through a database.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the gate fault detection method according to claim 1 are implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the gate fault detection method according to claim 1 are implemented.

Images