Real-time online monitoring method for pneumatic control valve

Abstract

The invention relates to the technical field of regulating valves, in particular to a real-time online monitoring method for a pneumatic regulating valve. The method comprises the following steps: firstly, collecting operation data of the pneumatic control valve, and transmitting the operation data to an edge computing node for preprocessing and lightweight processing to generate spectrum data; secondly, feature data in the frequency spectrum data are extracted and compressed, and regulating valve feature data are formed; and performing multi-source data fusion and dimension reduction on the characteristic data of the regulating valve at the central node to generate principal component fusion characteristics. And then key features in the fused features are extracted to generate a fault feature set, and a random forest model is adopted to obtain a fault identification result. Adjusting and updating the random forest model by adopting a self-adaptive model updating mechanism; and finally, performing extended training and optimization on the random forest model based on the historical operation data of the pneumatic control valve, and improving the multi-fault identification capability. According to the method, edge lightweight data processing is combined with deep learning, so that the accuracy, robustness and real-time performance of real-time monitoring of the pneumatic control valve are improved.

Description

Technical Field

[0001] The invention relates to the technical field of online monitoring of pneumatic control valves, and in particular to a real-time online monitoring method of pneumatic control valves. Background Art

[0002] Pneumatic control valve is an automatic control device that is driven by compressed air to adjust the flow, pressure, temperature or liquid level of the fluid. Its principle is to receive the pneumatic signal of the control system through the pneumatic actuator, drive the valve core to move or rotate in the valve body, change the valve opening, and thus control the flow of the fluid. Pneumatic control valves are widely used in the petroleum, chemical power, food, and pharmaceutical industries for precise regulation in production process control due to their advantages such as fast response, high precision, and good adaptability to harsh environments.

[0003] Real-time monitoring of pneumatic control valves can detect potential faults in a timely manner, avoid unexpected accidents, reduce downtime and reduce maintenance costs. At the same time, real-time monitoring helps to accurately control the operating status of the valve and extend the life of the pneumatic control valve equipment. The existing technology mainly collects key operating parameters through sensors, and uses signal processing technology and machine learning methods to analyze the data to achieve real-time monitoring of pneumatic control valves. This method can continuously monitor the operating status of pneumatic control valves and achieve high fault diagnosis accuracy and low missed diagnosis rate, but there are still some shortcomings.

[0004] Existing technologies usually detect and identify faults for specific types of control valves or single fault modes by integrating the collected data into a central processing terminal for processing and analysis. In actual pneumatic control valve application scenarios, multiple types of control valves are usually used in combination and multiple types of faults occur simultaneously. Under such complex working conditions, the data calculation complexity of existing technologies during real-time monitoring increases, and the processing difficulty increases. These problems result in the accuracy, robustness and real-time performance of real-time monitoring of pneumatic control valves still need to be improved.

[0005] Therefore, a real-time online monitoring method for pneumatic control valves is proposed. Summary of the invention

[0006] The purpose of the present invention is to provide a real-time online monitoring method for a pneumatic control valve. First, the operating data of the pneumatic control valve is collected and transmitted to the edge computing node for preprocessing and lightweight processing to generate spectrum data. Next, the spectrum data is feature calculated, and the feature data is extracted and compressed. At the central node, multi-source data fusion and dimensionality reduction are performed on the feature data to generate principal component fusion features. Then, key features are extracted from the fused features, a fault feature set is generated to obtain a fault identification model, and an adaptive model update mechanism is implemented to adjust and update the fault identification model; finally, the fault identification model is trained and optimized based on the historical operating data of the pneumatic control valve combined with a deep learning method to improve the multi-fault identification capability. The present invention improves the accuracy, robustness and real-time performance of real-time monitoring of pneumatic control valves by combining edge lightweight data processing with deep learning.

[0007] To achieve the above object, the present invention provides a real-time online monitoring method for a pneumatic control valve, comprising:

[0008] Collect real-time signal data during the operation of the pneumatic control valve and transmit it to the edge computing node for processing; the real-time signal data includes air chamber pressure data, valve stem displacement data, temperature data and vibration data;

[0009] Preprocessing and lightweight processing are performed on the real-time signal data to generate frequency spectrum data of the regulating valve;

[0010] Extracting key frequency components from the frequency spectrum data of the regulating valve by feature calculation, generating regulating valve feature data, and transmitting the data to a central computing node;

[0011] Extracting and reducing the dimension of the regulating valve characteristic data to obtain the principal component characteristics;

[0012] Performing multi-source data fusion on the principal component features to generate principal component fusion features;

[0013] Analyzing the fusion features and extracting key features to generate a control valve fault feature set;

[0014] The random forest model is used to classify and identify the control valve fault feature set to obtain the fault identification result;

[0015] Adopting an adaptive model update mechanism to adjust and update the random forest model;

[0016] The historical operation data of the pneumatic control valve and the fault identification results are collected, and the random forest model is extended and optimized to obtain a multi-fault identification model.

[0017] Further, the preprocessing includes low-pass filtering, baseline correction and outlier filtering;

[0018] x final (t) = F outlier (F base (F filt (x(t))));

[0019] Among them, x(t) is the signal data at time t, x final (t) is the preprocessed data signal, F filt (·) is the low-pass filtering operation, F base (·) is the baseline correction operation, F Outlier (·) is the outlier detection filtering operation.

[0020] Furthermore, the generation process of the frequency spectrum data of the regulating valve is completed by Fourier transform;

[0021] The preprocessed data signal is converted into a frequency domain signal by fast Fourier transform to generate the frequency spectrum data of the regulating valve; the conversion formula of the frequency domain signal is:

[0022]

[0023] Wherein, X(f) is the frequency spectrum data of the regulating valve, j is the imaginary unit, f is the frequency component, and x final (i) is the preprocessed data signal, and N is the total number of data points.

[0024] Further, feature calculation is performed on the frequency spectrum data of the regulating valve, including feature extraction, real-time analysis and compression;

[0025] By f max = argmax f |X(f)| Extract the maximum frequency f corresponding to the highest amplitude from the frequency spectrum data of the regulating valve max ; Through PSD(f) = |X(f)| 2 / N calculates the power spectrum density PSD(f) of the frequency spectrum data of the regulating valve; by Calculate the root mean square value RMS of the frequency spectrum data of the regulating valve;

[0026] For the maximum frequency f max The root mean square value (RMS) is used for abnormal monitoring. When the detected feature value exceeds the normal range, it is marked as abnormal.

[0027] The maximum frequency, RMS value and power spectrum density are calculated using the function X edge =compress(f max ,PSD(f),RMS) is compressed to obtain the regulating valve characteristic data; the regulating valve characteristic data is transmitted to the central computing node for processing.

[0028] Furthermore, the multi-source data fusion step includes principal component analysis and weighted fusion;

[0029] The regulating valve feature data from different edge computing nodes are reduced in dimension using a principal component analysis method to generate principal component features;

[0030] Z=A·[P edge (f),d edge (f),T edge (f),a edge (f)]

[0031] Among them, Z is the principal component feature, A is the eigenvector matrix of principal component analysis, P edge (f), d edge (f), T edge (f) and a edge (f) Characteristic data of pressure, displacement, temperature and vibration respectively;

[0032] The principal component features are weightedly fused according to the weights to generate principal component fusion features;

[0033] F=ω P ·Z P +ω d ·Z d +ω T ·Z T +ω a ·Z a ;

[0034] Among them, Z P , Z d , Z T and Z a are the principal component characteristics of pressure, displacement, temperature and vibration, ω P ,ω d ,ω T and ω a are the weights of the corresponding principal component features.

[0035] Furthermore, the steps of acquiring the fault feature set are:

[0036] Analyze the fusion features and use F fault = E(F) extracts key features related to the fault and generates a fault feature set F fault , where E(·) is the feature extraction function.

[0037] Furthermore, the step of acquiring the fault identification model includes:

[0038] Using random forest C fault =RF(F fault) classify the fault feature set and output the fault identification result; C fault is the fault identification result, and RF(·) is the random forest model.

[0039] Furthermore, the steps of the adaptive update mechanism are:

[0040] Calculating the difference between the fault identification result and the actual fault type to determine the triggering condition of the adaptive mechanism;

[0041] Collect the latest operating data of the pneumatic control valve to generate a real-time fault feature set;

[0042] The parameters of the random forest model are updated using an adaptive online learning algorithm according to the real-time fault feature set.

[0043] θ RF,new =θ RF,ori -η RF Δ θRF Φ(F new, ,C fault );

[0044] Among them, θ RF,ori and θ RF,new are the random forest model parameters before and after updating respectively; η RF is the learning rate, Φ(F new, ,C fault ) is the loss function, Δ θRF is the rate of change of the loss function, F new, It is a real-time fault feature set.

[0045] Furthermore, the training process of the multi-fault identification model is:

[0046] Collect historical data of the pneumatic control valve, including historical characteristic data and the corresponding fault identification results.

[0047] Performing extended training on the random forest model to extract correlation features between historical feature data and fault identification results;

[0048]

[0049] Among them, DNN multi is the correlation feature, FC(·) is a three-layer convolution operation, x' history (m) is the mth historical feature, C history (p) is the p-th fault identification result.

[0050] The extended training process includes:

[0051] Input the historical feature data and the corresponding fault identification result into a three-layer convolution layer, extract fault correlation features, and output a first correlation feature map;

[0052] Using maximum pooling to reduce the dimension of the fault correlation features, reducing the size of the first correlation feature map and retaining important information, and outputting a second correlation feature map;

[0053] Expand the second correlation feature map into a vector and extract the fault correlation global feature;

[0054] The fault-related global features are converted into probability distribution of fault types through a softmax activation function to obtain the related features;

[0055] By using the associated features, the random forest model is adjusted to obtain a multi-fault identification model;

[0056] Furthermore, the optimization process of the multi-fault identification model is:

[0057] Collect the latest operating data of the pneumatic control valve, and use the multi-fault identification model to obtain a multi-fault identification result;

[0058] Calculating the difference between the multiple fault identification result and the actual multiple fault type in real time, and optimizing and adjusting the multiple fault identification model according to the difference;

[0059]

[0060] in, Parameters of the preliminary model for multiple fault identification, η RF is the learning rate, Φ(F train ,C multi ) is the loss function, is the rate of change of the loss function, F train is the fault feature set of historical operation data, C multi These are the real multiple fault types corresponding to the historical operation data.

[0061] Compared with the prior art, the present invention has the following beneficial effects:

[0062] 1. The present invention first pre-processes the pneumatic control valve operation data, converts the pre-processed time domain signal into a frequency domain signal through fast Fourier transform, generates the control valve spectrum data, and extracts the key frequency components from it. Through this step, the method converts the complex time domain signal into a frequency domain signal that is easier to analyze, extracts the key frequency information reflecting the operating status of the control valve, solves the problem of high computational complexity of time domain signal analysis in the prior art, and provides data support for improving the real-time and accuracy of online monitoring of pneumatic control valves.

[0063] 2. The present invention transfers the data feature extraction task to the edge computing node, detects abnormal features and marks abnormal values ​​by extracting key feature data and analyzing and monitoring in real time. After compression and dimension reduction processing, the key feature data is fused to generate principal component fusion features through multi-source data fusion. This method solves the computational compliance problem of centralized data processing in the prior art and improves data processing efficiency. At the same time, multi-source data fusion simplifies the data structure, enhances the processing and fault detection capabilities of multi-source information, and improves the robustness and real-time performance of online monitoring of pneumatic control valves.

[0064] 3. The present invention generates a fault feature set by extracting key features related to the fault from the principal component fusion features; then uses a random forest model to classify and identify the fault feature set to obtain fault identification results; then adjusts the model parameters in real time through an adaptive mechanism, and based on historical operating data and fault identification results, performs extended training on the optimized random forest model, and finally adjusts and optimizes the model through a cross entropy loss function to obtain a multi-fault identification model. This method solves the problem that the prior art is difficult to identify concurrent faults by combining fault feature extraction with random forest classification. By combining adaptive updates with deep learning training, the accuracy of real-time monitoring of pneumatic control valves is comprehensively improved. BRIEF DESCRIPTION OF THE DRAWINGS

[0065] Figure 1 A schematic flow chart of a real-time online monitoring method for a pneumatic control valve provided by an embodiment of the present invention;

[0066] Figure 2 A schematic diagram of a flow chart of obtaining characteristic data of a regulating valve provided in an embodiment of the present invention;

[0067] Figure 3 A schematic diagram of the training optimization process of the multi-fault identification model provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0068] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0069] The present invention provides a real-time online monitoring method for a pneumatic control valve. Figure 1 .

[0070] Embodiment 1

[0071] As an embodiment of the present invention, refer to Figure 1S10 in the system is used to collect real-time data during the operation of pneumatic control valves, including air chamber pressure data, valve stem displacement data, temperature data and vibration data, and transmits them to edge nodes for pre-processing and lightweight preliminary processing to generate control valve spectrum data.

[0072] The edge node is deployed in the valve assembly area of ​​each pneumatic control valve, and each edge node is configured with an edge computing device; the pneumatic control valve operation data is collected by a sensor and transmitted to the edge computing device through a wired device;

[0073] The preprocessing includes low-pass filtering, baseline correction and outlier detection filtering; for the original data signal, high-frequency noise is first removed by a low-pass filter;

[0074]

[0075] x filtered (t) is the filtered signal, h k is the filter coefficient, N is the filter order, and x(t) is the original data signal of the pneumatic control valve.

[0076] The baseline drift in the data signal is eliminated by the moving average method, and the corrected signal is:

[0077]

[0078] x corrected (t) is the signal after baseline correction, and M is the window size of the moving average.

[0079] The 3-times standard deviation method was used to identify and exclude outliers;

[0080]

[0081] x final (t) is the time domain signal after preprocessing, μ x is the average signal after baseline correction, σ x Standard deviation of the signal after baseline correction, NA means outliers have been filtered.

[0082] This embodiment reduces the impact of noise, drift and abnormal data caused by non-faults in the preprocessing step through low-pass filtering, baseline correction and outlier detection and filtering, ensures the stability and reliability of the data, and provides high-quality input for subsequent analysis and processing.

[0083] The generation of the frequency spectrum data of the regulating valve mainly adopts the fast Fourier transform method;

[0084] The pre-processed time domain signal is converted into a frequency domain signal by fast Fourier transform to generate the frequency spectrum data of the regulating valve; the conversion formula of the frequency domain signal is:

[0085]

[0086] Wherein, X(f) is the frequency spectrum data of the regulating valve, j is the imaginary unit, f is the frequency component, and x final (i) is the preprocessed data signal, and N is the total number of data points.

[0087] This embodiment uses Fourier fast transform to convert time domain signals into frequency domain signals, making complex time domain data easier to analyze. In scenarios with obvious frequency characteristics, it not only improves the efficiency of data processing, but also reduces redundant analysis of raw data, providing support for improving the rapid response capability of real-time monitoring of pneumatic control valves.

[0088] Further, refer to Figure 1 S20 in which S20 is used to perform feature calculation on the frequency spectrum data of the regulating valve, further extract data features, and generate characteristic data of the regulating valve;

[0089] Reference Figure 2 ,The feature calculation process includes feature extraction, real-time analysis and compression;

[0090] By f max = argmax f |X(f)| Extract the maximum frequency f corresponding to the highest amplitude from the frequency spectrum data of the regulating valve max ;

[0091] Through PSD(f) = |X(f)| 2 / N calculates the power spectrum density PSD(f) of the frequency spectrum data of the regulating valve;

[0092] pass Calculate the root mean square value RMS of the frequency spectrum data of the regulating valve;

[0093] For the maximum frequency f max The root mean square value (RMS) is used for abnormal monitoring. When the detected feature value exceeds the normal range, it is marked as abnormal.

[0094] The maximum frequency, RMS value and power spectrum density are calculated using the function X edge =compress(f max ,PSD(f),RMS) is compressed to obtain the control valve characteristic data;

[0095] The regulating valve characteristic data is transmitted to a central computing node via a wireless device.

[0096] This embodiment improves the accuracy and real-time performance of data analysis by performing feature calculations on the frequency spectrum data of the control valve, including extracting the maximum power, power spectrum density and root mean square value, and performing real-time abnormality monitoring and data compression. This method can capture abnormal features in a timely manner to ensure the rapid identification of potential problems of the pneumatic control valve; at the same time, by compressing key feature data, it reduces the burden of storage and transmission, and also reduces the data complexity of subsequent analysis.

[0097] Further, refer to Figure 1 S30 in which S30 is used to extract the principal component of the regulating valve feature data from the edge computing node and reduce the dimension, and then perform multi-source data fusion to generate the principal component fusion feature, including principal component analysis and weighted fusion;

[0098] The control valve feature data from the edge computing node is reduced in dimension using the principal component analysis method to generate principal component features;

[0099] Z=A·[P edge (f),d edge (f),T edge (f),a edge (f)]

[0100] Among them, Z is the principal component feature, A is the eigenvector matrix of principal component analysis, P edge (f), d edge (f), T edge (f) and a edge (f) Characteristic data of air chamber pressure, valve stem displacement, temperature and vibration respectively;

[0101] The principal component features are weighted and fused according to the weights to generate principal component fusion features;

[0102] F=ω P ·Z P +ω d ·Z d +ω T ·Z T +ω a ·Z a ;

[0103] Among them, Z P , Z d , Z T and Z a are the principal component characteristics of pressure, displacement, temperature and vibration, ω P ,ω d ,ω T and ω a are the weights of the corresponding principal component features, respectively. The weights are set by professionals according to the actual application scenarios.

[0104] This embodiment extracts the most important principal component features by performing principal component analysis on multi-source data signals such as pressure, displacement, temperature, and vibration, and then performs weighted fusion on different signals, which simplifies the data dimension, improves the efficiency and accuracy of signal processing, and retains key fault feature information, which helps to improve the accuracy of fault identification.

[0105] Further, refer to Figure 1 S40 in which S40 is used to analyze the principal component fusion features and extract key features to generate a control valve fault feature set;

[0106] A random forest model is used to classify and identify the control valve fault feature set to obtain a fault identification model;

[0107] The steps of obtaining the fault feature set are:

[0108] According to the failure mode of the pneumatic control valve, the fusion characteristics are analyzed and the F fault =E(F) extracts key features related to the fault;

[0109] The fault types and key data characteristics are listed in Table 1;

[0110] Table 1 Fault types and related data characteristics

[0111]

[0112] Vibration fault characteristics: by detecting abnormal frequency fluctuations or high-frequency components in the vibration signal;

[0113] Pressure fault characteristics: Analyze the maximum and minimum values ​​of the air chamber pressure signal in the time domain to identify abnormal air chamber pressure fluctuations;

[0114] Temperature fault characteristics: Analyze the time domain characteristics of the temperature signal to determine whether the temperature exceeds the safe range;

[0115] Displacement fault characteristics: monitor displacement signal fluctuations through the time domain and frequency domain characteristics of the displacement signal;

[0116] Signal Rate of Change: Monitors rapid changes in a signal.

[0117] Generate fault signature set F fault =[f max ,f min ,f mean ,f rms ,f freq ,f psd ,f rate ], f max and f min are the peak and minimum values ​​of various signals, respectively, fmean and f rms are the mean and RMS values ​​of various signals, respectively, reflecting the overall trend or steady state of the signal; f freq is the main frequency component used to identify periodic faults; f psd is the signal power spectrum density, which reflects the energy distribution of the signal at different frequencies; f rate is the rate of signal change used to identify sudden signal changes.

[0118] This embodiment generates a fault feature set of a pneumatic control valve by fusing the time domain and frequency domain features of multi-source data signals of pressure, displacement, temperature and vibration, covering various key operating parameters of the pneumatic control valve, and providing rich features for the establishment of subsequent fault identification models. This method extracts fault-related features, removes redundant or useless data, simplifies the data analysis process, and improves the efficiency of subsequent fault identification.

[0119] The step of acquiring the fault identification model includes:

[0120] According to historical experience, a random forest model is constructed and the fault feature set F fault Input into the random forest model; each decision tree in the random forest is trained on a different subset of the fault feature set to form independent classification rules for identifying different fault modes;

[0121] Each decision tree in the random forest independently classifies and predicts the fault feature set, and each tree predicts the fault type corresponding to the current feature based on its classification rules;

[0122]

[0123] Where T i represents the i-th decision tree, is the classification result of the fault type by the i-th decision tree.

[0124] Random forest uses a voting mechanism to make the final judgment on the fault type and generate fault identification results.

[0125]

[0126] Among them, mode is the majority vote mechanism of random forest.

[0127] This embodiment integrates multiple decision trees through the random forest model to classify and identify faults, and adopts the majority vote mechanism to improve the accuracy and robustness of fault identification. This method can handle complex multi-dimensional features, quickly classify and accurately identify fault features, and ensure real-time and high efficiency while improving fault detection accuracy.

[0128] Further, refer to Figure 1S50 in which S50 is used to adjust and update the random forest model by using an adaptive model update mechanism;

[0129] The steps of the adaptive update mechanism are: regularly collecting the latest operating data of the pneumatic control valve, generating a real-time fault feature set F according to known feature extraction rules, new ;

[0130] Use an adaptive online learning algorithm to update the parameters of the fault identification model based on the real-time fault feature set;

[0131] Adjusting the parameters of the fault identification model in real time and updating the model through online learning;

[0132]

[0133] Among them, θ RF,ori and θ RF,new are the random forest model parameters before and after updating respectively; η RF is the learning rate, which is used to control the step size of model parameter adjustment; Φ(F new, ,C fault ) is the loss function; is the gradient of the loss function to the model parameters, indicating the direction of parameter adjustment; F new It is a real-time fault feature set.

[0134] The loss function Φ(·) reflects the model classification error. The loss function is calculated based on the difference between the real-time data and the recognition result;

[0135]

[0136] Among them C q is the label of the actual fault type, c q The fault types identified by the fault identification model;

[0137] When Φ(F new ,C fault-realtime ) value increases, the adaptive mechanism will be automatically triggered to adjust the fault identification model by introducing new feature data.

[0138] This embodiment dynamically adjusts the fault identification model through an adaptive update mechanism, so that the model can be continuously optimized, and the accuracy and robustness of fault type classification and identification can be improved. This method can adapt to changes in the operating environment of the pneumatic control valve, reduce false positives and false negatives of faults, and ensure that the model maintains efficient fault detection capabilities in long-term operation.

[0139] Further, refer to Figure 1S60 in the example is used to collect historical data of the pneumatic control valve, and the random forest model is extended and optimized to obtain a multi-fault recognition model.

[0140] Reference Figure 3 , the training process of the multi-fault identification model is:

[0141] Collect historical operating data of pneumatic control valves D history =(X history ,C history ), X history is the historical feature data set of each signal, C history is the fault identification result set;

[0142] Performing extended training on the random forest model to extract correlation features between historical feature data and fault identification results;

[0143]

[0144] Among them, DNN multi is the correlation feature, FC(·) is a three-layer convolution operation, x' history (m) is the mth historical feature, C history (p) is the p-th fault identification result.

[0145] The specific training process is:

[0146] The historical feature data and the corresponding fault recognition results are input into a three-layer convolutional layer, using Z conv =σ(W conv *X history +b conv ) extract fault correlation features and output a first correlation feature graph; Z conv is the first correlation feature map, W conv is the convolution kernel weight, b conv is the bias, * represents the convolution operation, and σ(·) is the activation function

[0147] Use maximum pooling to perform the first correlation feature map Z conv Perform dimensionality reduction to reduce the first correlation feature map Z conv The size of the image is 0.0447×0.08 and important information is retained, and the second correlation feature map is output;

[0148] Expand the second associated feature map into a vector and pass it through Z fc =σ(W fc ·Z pool +b fc ) extracts the global features of fault association; where Z fc is the global feature of fault association, W fc is the weight, Zpool is the second correlation feature map, b fc For bias.

[0149] The fault-related global features are converted into probability distribution of fault types through a softmax activation function to obtain the related features;

[0150] By using the associated features, the random forest model is adjusted to obtain a multi-fault identification model;

[0151] This embodiment uses a deep neural network to expand the training of the fault identification model, which can improve the accuracy of fault classification and the ability to identify multiple faults. By using a deep neural network to extract complex features from multi-dimensional signals, the model can effectively identify concurrent faults and ensure efficient operation under complex working conditions.

[0152] The optimization process of the multi-fault identification model is:

[0153] Collect the latest operating data of the pneumatic control valve, and use the multi-fault identification model to obtain a multi-fault identification result;

[0154] A cross entropy loss function is used to calculate the difference between the multi-fault identification result and the actual multi-fault type in real time, and the multi-fault identification model is optimized and adjusted according to the difference;

[0155]

[0156] in, is the parameter of the multi-fault identification model, η RF is the learning rate, Φ(F train ,C multi ) is the loss function, is the rate of change of the loss function, F train is the fault feature set of historical operation data, C multi These are the real multiple fault types corresponding to the historical operation data.

[0157] This embodiment improves the fault detection accuracy and robustness of the multi-fault identification model by optimizing the loss function and updating the model parameters. Through optimization, the model can adapt to complex working conditions of concurrent faults, improve the real-time performance and detection efficiency of pneumatic control valve fault detection, and ensure rapid response and efficient handling of multiple faults.

[0158] This embodiment improves the fault detection and identification capabilities of pneumatic control valves by combining edge computing with a variety of data processing methods. Extract key features from the multi-dimensional signals of the pneumatic control valve operation data to accurately identify a variety of fault types. The adaptive mechanism ensures that the model is continuously optimized during operation, improves the robustness and accuracy of fault detection, and reduces false positives and missed reports. Edge computing and lightweight processing reduce the time and storage costs of data processing, improve the real-time and efficiency of real-time monitoring of pneumatic control valves, and ensure timely response and fault handling. Through the training and optimization of the multi-fault identification model, the pneumatic control valve operation system has a high adaptability and can still work stably when concurrent faults occur in complex situations.

[0159] Embodiment 2

[0160] As an implementation mode of the present invention, this embodiment describes a specific implementation mode of a pneumatic control valve operation system of a chemical plant A.

[0161] Collect real-time data of the pneumatic control valve in Chemical Plant A during operation, including chamber pressure data, valve stem displacement data, temperature data, and vibration data. Each type of data is collected at 12 time points.

[0162] The sensor is transmitted to the edge node for pre-processing and lightweight preliminary processing to generate the control valve spectrum data;

[0163] The edge node is deployed in the valve assembly area of ​​each pneumatic control valve, and each edge node is configured with an edge computing device; the pneumatic control valve operation data is collected by a sensor and transmitted to the edge computing device through a wired device;

[0164] Some of the collected data are listed in Table 2;

[0165] Table 2 Example of data collection for Chemical Plant A

[0166]

[0167] For the original data signal, firstly, high-frequency noise is removed by low-pass filter; baseline drift in the data signal is eliminated by moving average method, and outliers are identified and excluded by 3 times standard deviation method;

[0168] The pre-processed time domain signal is converted into a frequency domain signal by fast Fourier transform to generate the frequency spectrum data of the regulating valve;

[0169] Performing feature calculation on the frequency spectrum data of the regulating valve, further extracting data features, and generating regulating valve feature data; the feature calculation process includes feature extraction, real-time analysis and compression;

[0170] Extract the maximum frequency corresponding to the highest amplitude from the frequency spectrum data of the regulating valve; calculate the power spectrum density of the frequency spectrum data of the regulating valve; calculate the root mean square value of the frequency spectrum data of the regulating valve;

[0171] The maximum frequency and RMS value are monitored for abnormalities. When the detected feature value exceeds the normal range, it is marked as abnormal.

[0172] The maximum frequency, root mean square value and power spectrum density are compressed using a function to obtain the characteristic data of the control valve and transmit it to the central computing node.

[0173] The control valve feature data from the edge computing node is reduced in dimension using the principal component analysis method to generate principal component features;

[0174] The principal component features are weighted and fused according to the weights to generate principal component fusion features;

[0175] Analyze the fusion features and extract key features to generate a control valve fault feature set; use a random forest model to classify and identify the control valve fault feature set to obtain a fault identification model;

[0176] The steps of acquiring the fault feature set are: analyzing the fusion features according to the known fault modes of the pneumatic control valve, and generating the fault feature set by extracting the key features related to the fault.

[0177] Vibration fault characteristics: by detecting abnormal frequency fluctuations or high-frequency components in the vibration signal;

[0178] Pressure fault characteristics: Analyze the maximum and minimum values ​​of the air chamber pressure signal in the time domain to identify abnormal air chamber pressure fluctuations;

[0179] Temperature fault characteristics: Analyze the time domain characteristics of the temperature signal to determine whether the temperature exceeds the safe range;

[0180] Displacement fault characteristics: monitor displacement signal fluctuations through the time domain and frequency domain characteristics of the displacement signal;

[0181] Signal Rate of Change: Monitors rapid changes in a signal.

[0182] A random forest model is constructed based on historical experience, and the fault feature set is input into the random forest model. Each decision tree in the random forest is trained based on a different subset of the fault feature set to form independent classification rules for identifying different fault modes.

[0183] Each decision tree in the random forest independently classifies and predicts the fault feature set, and each tree predicts the fault type corresponding to the current feature based on its classification rules;

[0184] Random forest uses a voting mechanism to make the final judgment on the fault type and finally generate the fault identification result.

[0185] Implementing an adaptive model update mechanism to adjust and update the fault identification model;

[0186] The steps of the adaptive update mechanism are: regularly collecting the latest operating data of the pneumatic control valve and generating a real-time fault feature set according to known feature extraction rules;

[0187] Using an adaptive online learning algorithm, updating the parameters of the fault identification model according to the real-time fault feature set; adjusting the parameters of the fault identification model in real time, and updating the model through online learning;

[0188] The loss function reflects the model classification error. The loss function is calculated based on the difference between the real-time data and the recognition result. When the loss function increases, the adaptive mechanism is automatically triggered to adjust the fault recognition model by introducing new feature data.

[0189] The historical operation data of the pneumatic control valve is collected, and the fault identification model is trained and optimized using a deep learning model to obtain a multi-fault identification model.

[0190] The training process of the multi-fault identification model is:

[0191] Collect historical operating data of pneumatic control valves, including historical feature data and feature data output by the fault identification model corresponding to fault labels; after preprocessing, input into a deep neural network to extract the relationship features between operating data and fault types;

[0192] The optimization process of the multi-fault identification model is:

[0193] The cross entropy loss function is used to optimize the preliminary model for multiple fault identification;

[0194] According to the above steps, the data collected by Chemical Plant A in Table 2 are processed, and the multiple fault types of the pneumatic control valve operation system are obtained, which are air chamber leakage, valve stem sticking or actuator abnormality, valve overheating and mechanical looseness.

[0195] This embodiment performs outlier detection, low-pass filtering and baseline correction on the collected air chamber pressure, valve stem displacement, temperature and vibration signals through a preprocessing step to ensure the quality and accuracy of the data, laying the foundation for subsequent analysis; then, edge computing and lightweight data processing technology are used to quickly analyze and process the data, reducing the computing load and space occupancy, and providing data support for improving the real-time and efficiency of fault detection; then, through feature extraction of multi-source data signals, the operating status of the pneumatic control valve equipment can be fully captured, and various fault types of the pneumatic control valve can be identified.

[0196] In the fault identification stage, this embodiment uses deep learning and random forest models to accurately classify and identify complex concurrent faults, ensuring efficient processing of different types of fault signals; through the adaptive mechanism, the fault identification model can automatically adjust the model parameters according to the changes in the actual operation data, ensuring the robustness in long-term operation and the continuous optimization of fault identification.

[0197] This embodiment provides data, theoretical and technical support for real-time online monitoring of pneumatic control valves through multiple steps such as lightweight processing, edge computing, key feature extraction, and fault identification, thereby improving the accuracy, real-time performance and robustness of real-time online monitoring of pneumatic control valves.

[0198] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A real-time online monitoring method for a pneumatic control valve, characterized in that: include: Collect real-time signal data during the operation of the pneumatic control valve and transmit it to the edge computing node for processing; The real-time signal data includes air chamber pressure data, valve stem displacement data, temperature data and vibration data; Preprocessing and lightweight processing are performed on the real-time signal data to generate frequency spectrum data of the regulating valve; Extracting key frequency components from the frequency spectrum data of the regulating valve by feature calculation, generating regulating valve feature data, and transmitting the data to a central computing node; Extracting and reducing the dimension of the regulating valve characteristic data to obtain the principal component characteristics; Performing multi-source data fusion on the principal component features to generate principal component fusion features; Analyzing the fusion features and extracting key features to generate a control valve fault feature set; The random forest model is used to classify and identify the control valve fault feature set to obtain the fault identification result; Adopting an adaptive model update mechanism to adjust and update the random forest model; The historical operation data of the pneumatic control valve and the fault identification results are collected, and the random forest model is extended and optimized to obtain a multi-fault identification model.

2. A real-time online monitoring method for a pneumatic control valve according to claim 1, characterized in that: The preprocessing includes low-pass filtering, baseline correction and outlier detection filtering; Removing high-frequency noise from the real-time signal data by using a low-pass filter; The moving average method is used to eliminate the baseline drift in the data signal and correct the data signal; Identify and exclude outliers in data signals using the three-times standard deviation method.

3. The real-time online monitoring method of a pneumatic control valve according to claim 1 is characterized in that: The generation process of the regulating valve spectrum data is as follows: The pre-processed real-time signal data is converted into a frequency domain signal by fast Fourier transform to generate the regulating valve spectrum data.

4. The real-time online monitoring method of a pneumatic control valve according to claim 1 is characterized in that: The process of acquiring the characteristic data of the regulating valve includes: Extract the maximum frequency corresponding to the highest amplitude from the frequency spectrum data of the regulating valve; calculate the power spectrum density of the frequency spectrum data of the regulating valve; calculate the root mean square value of the frequency spectrum data of the regulating valve; Performing abnormal monitoring on the maximum frequency and the root mean square value, and marking them as abnormal when the detected maximum frequency and the root mean square value are beyond a normal range; The maximum frequency, the root mean square value and the power spectrum density are compressed using a compression function to obtain the regulating valve characteristic data, and are transmitted to a central computing node.

5. The real-time online monitoring method of a pneumatic control valve according to claim 1, characterized in that: The multi-source data fusion step includes principal component analysis and weighted fusion; The regulating valve characteristic data includes air chamber pressure characteristics, valve stem displacement characteristics, temperature characteristics and vibration characteristics; Extracting principal components from the regulating valve characteristic data using principal component analysis and reducing the dimension to generate principal component features; The principal component features are weighted and fused to generate principal component fusion features.

6. A real-time online monitoring method for a pneumatic control valve according to claim 1, characterized in that: The steps of obtaining the fault feature set are: Extract key features related to the fault from the principal component fusion features through a feature extraction function to generate a fault feature set; The fault feature set includes vibration fault features, pressure fault features, temperature fault features, displacement fault features and signal change rate.

7. A real-time online monitoring method for a pneumatic control valve according to claim 1, characterized in that: The step of obtaining the fault identification result includes: Constructing a random forest model, and inputting the fault feature set into the random forest model; The fault feature set is divided into several training subsets for training, and independent classification rules are formed for the fault modes; Classify and predict the fault feature set; wherein each tree in the random forest predicts the fault type corresponding to the current feature according to the corresponding independent classification rule; The fault identification result is obtained by adopting a majority vote mechanism.

8. The real-time online monitoring method of a pneumatic control valve according to claim 1, characterized in that: The steps of the adaptive update mechanism are: Calculating the difference between the fault identification result and the actual fault type to determine the triggering condition of the adaptive mechanism; Collect the latest operating data of the pneumatic control valve to generate a real-time fault feature set; The parameters of the random forest model are updated using an adaptive online learning algorithm according to the real-time fault feature set.

9. The real-time online monitoring method of a pneumatic control valve according to claim 1, characterized in that: The training process of the multi-fault identification model is: Collecting historical data of the pneumatic control valve, including historical characteristic data and the corresponding fault identification results; Performing the extended training on the random forest model to extract correlation features between the historical feature data and the fault identification result; The extended training process includes: Input the historical feature data and the corresponding fault identification result into a three-layer convolution layer, extract fault correlation features, and output a first correlation feature map; Using maximum pooling to reduce the dimension of the first correlation feature map, reducing the size of the first correlation feature map and retaining important information, and outputting a second correlation feature map; Expand the second correlation feature map into a vector and extract the fault correlation global feature; The fault-related global features are converted into probability distribution of fault types through a softmax activation function to obtain the related features; The random forest model is adjusted through the associated features to obtain a multi-fault identification model.

10. The real-time online monitoring method of a pneumatic control valve according to claim 1, characterized in that: The optimization process of the multi-fault identification model is: Collect the latest operating data of the pneumatic control valve, and use the multi-fault identification model to obtain a multi-fault identification result; A cross entropy loss function is used to calculate in real time the difference between the multi-fault identification result and the actual multi-fault type, and the multi-fault identification model is optimized and adjusted according to the difference.

Images