CN117470529A - A process control system valve static friction fault detection method and system - Google Patents

Abstract

The invention discloses a method and a system for detecting static friction faults of a valve of a process control system, comprising the following steps: acquiring controller output data and controlled process variable data of a valve; extracting slow features from the controller output data and the controlled process variable data; calculating the Hurst index of the slowest slow features in the slow features; determining a valve static friction detection index according to the Hurst index; judging whether the valve has static friction faults according to the static friction detection index of the valve, and obtaining a fault detection result. The noise robustness can be effectively enhanced, and meanwhile, the detection performance is improved, and particularly, the valve static friction-caused aperiodic random oscillation has good detection performance.

Description

A process control system valve static friction fault detection method and system

Technical field

The invention relates to the technical field of valve static friction fault detection, and in particular to a method and system for detecting valve static friction fault in a process control system.

Background technique

The statements in this section merely provide background technical information related to the present invention and do not necessarily constitute prior art.

Control valves are widely used in process control systems to regulate fluid flow, pressure and temperature in various industrial processes, such as oil and gas, construction, chemical, petrochemical, power and water treatment production processes. They are key execution equipment for achieving precise control of the production process and maintaining product quality. However, one of the challenges in controlling the normal operation of valves is the occurrence of valve static friction, which can cause oscillation of the control loop, thus reducing its control performance and product quality. It also accelerates the wear and aging of valves and related equipment, causing system failure and even causing Production safety accidents. According to surveys, 20%-30% of control loops will cause oscillation in the process control loop due to valve stiction. Therefore, studying valve static friction fault detection methods is of great significance in improving process control system reliability, control performance and product quality.

The inventor found that valve oscillation caused the process data to oscillate, with significant temporal dynamic characteristics and nonlinear characteristics, which caused significant changes in the long-term correlation information in the process data, and the existing valve static friction detection technology did not fully utilize it. Time-series dynamic information and long-term correlation information. At the same time, the existing technology has the disadvantage of poor robustness to random noise in the data, which makes the detection and identification accuracy of valve static friction faults low.

Contents of the invention

In order to solve the above problems, the present invention proposes a process control system valve static friction fault detection method and system, which improves the detection accuracy of valve static friction fault.

In order to achieve the above objects, the present invention adopts the following technical solutions:

In the first aspect, a method for detecting static friction faults of valves in process control systems is proposed, including:

Obtain controller output data and controlled process variable data of the valve;

Extract slow features from controller output data and controlled process variable data;

Calculate the Hurst index of the slowest feature among the slow features;

According to the Hurst index, determine the valve static friction detection index;

According to the valve static friction detection index, it is judged whether the valve has static friction failure and the fault detection result is obtained.

In the second aspect, a process control system valve static friction fault detection system is proposed, including:

Data acquisition module, used to obtain the controller output data and controlled process variable data of the valve;

Slow feature extraction module, used to extract slow features from controller output data and controlled process variable data;

The Hurst index calculation module is used to calculate the Hurst index of the slowest feature among the slow features;

The valve static friction detection index acquisition module is used to determine the valve static friction detection index based on the Hurst index;

The fault detection result acquisition module is used to determine whether a static friction fault occurs in the valve according to the valve static friction detection index and obtain the fault detection result.

In a third aspect, an electronic device is proposed, including a memory and a processor, and computer instructions stored in the memory and run on the processor. When the computer instructions are run by the processor, a process control system valve stiction failure is completed. The steps described in the detection method.

In a fourth aspect, a computer-readable storage medium is proposed for storing computer instructions. When the computer instructions are executed by a processor, the steps described in a process control system valve stiction fault detection method are completed.

Compared with the prior art, the beneficial effects of the present invention are:

1. The present invention obtains the temporal dynamic characteristics of the acquired data by extracting the slow characteristics of the acquired data. By calculating the Hurst index of the slowest characteristic, the long-term correlation information of the data can be obtained. The Hurst index is used to determine the valve static friction detection index, and the Hurst index is used to determine the valve static friction detection index. The valve static friction detection index determines whether the valve has static friction faults, which improves the detection accuracy of valve static friction faults, especially the detection performance when static friction causes irregular oscillations in the control loop.

2. When extracting slow features, the present invention simultaneously eliminates noise signals and improves the robustness of the detection method to noise signals.

Additional advantages of the invention will be set forth in part in the description which follows, and in part will be obvious from the description, or may be learned by practice of the invention.

Description of the drawings

The description and drawings that constitute a part of this application are used to provide a further understanding of this application. The illustrative embodiments and their descriptions of this application are used to explain this application and do not constitute an improper limitation of this application.

Figure 1 is a flow chart of the method disclosed in Embodiment 1;

Figure 2 is a typical control loop structure diagram of the valve;

Figure 3 is a flow chart of data collection and preprocessing disclosed in Embodiment 1;

Figure 4 is a flow chart of slow feature extraction and denoising based on SFA disclosed in Embodiment 1;

Detailed ways

The present invention will be further described below in conjunction with the accompanying drawings and examples.

It should be noted that the following detailed description is illustrative and is intended to provide further explanation of the present application. Unless otherwise defined, all technical and scientific terms used herein have the same meanings commonly understood by one of ordinary skill in the art to which this application belongs.

Example 1

The typical valve control loop structure is shown in Figure 2. The valve is controlled based on the set value SP controller output (ControllerOutput, OP) data, and the controlled process variable (PV) after the valve is controlled is obtained through the sensor. )data.

When a static friction failure occurs in a valve, it often causes the OP data and PV data in the loop to oscillate, thereby affecting the stability and safety of the control system, resulting in reduced product quality, increased energy consumption and equipment wear. Valve static friction causes obvious dynamic changes and nonlinear characteristics of data in the process loop. The SFA method has the ability to extract dynamic features and denoise time series data. The Hurst index can calculate the long-term correlation of time series data and is often used. Detection of nonlinear characteristics. Therefore, this embodiment proposes a process control system valve static friction fault detection method. This method is based on the double-layer ML method and combines the SFA and Hurst index methods to extract dynamic characteristics and long-term correlation information in the reconstructed expanded-dimensional data matrix data. Among them, data-based reconstruction and dimensionality expansion help to make full use of the phase difference information and oscillation information in OP and PV data; the SFA-based method can extract the slowly changing time series dynamic characteristics information in the reconstructed dimensionality expansion data; based on Hurst The exponential method can calculate long-term correlation information in time series data to quantify the nonlinear characteristics of the data, thereby improving the accuracy and sensitivity of valve stiction fault detection.

As shown in Figure 1, this embodiment discloses a process control system valve static friction fault detection method, including:

S1: Obtain the controller output data and controlled process variable data of the valve.

Obtain the controller output data (OP data) and controlled process variable data (PV data) of the valve in the process control system.

S2: Extract slow features from controller output data and controlled process variable data.

In this embodiment, the controller output data and the controlled process variable data are preprocessed to obtain preprocessed data; slow features are extracted from the preprocessed data.

As shown in Figure 3, the preprocessing process includes standardization, data reconstruction and dimensionality expansion.

Equation (1) is used to standardize the obtained OP data and PV data.

Among them, x(k) is the OP data and PV data obtained at time k, and σ are the mean and variance of data x(k), (k=1,...,N) respectively.

The data after normalization are reconstructed by making a difference between the OP data and PV data after normalization, and the reconstructed data d(k) is obtained.

d(k)＝ _vPV (k) _-vOP (k)(2)

Among them, v _PV (k) and v _OP (k) are the PV data and OP data after normalization respectively.

The reconstructed data d(k) is one-dimensional data. Use l lag samples to perform data dimension expansion on the reconstructed data d(k) to obtain the data matrix D _d (k), which is the preprocessed data.

Among them, l is the number of lagged samples.

The slow feature analysis algorithm (SFA method) is used to extract slow features from the preprocessed data, as shown in Figure 4. The specific process is:

S21: Standardize the preprocessed data D _d (k) to obtain the standardized matrix.

Use formula (1) to normalize D _d (k) to obtain the standardized matrix

S22: Perform singular value decomposition (SVD) on the covariance matrix of the standardized matrix, and calculate the spherical matrix Z.

S23: Difference matrix for spheroidized matrix Perform singular value decomposition and calculate the SFA transformation matrix W.

W＝PA ^-1/2 U ^T (7)

in, Z(k) and Z(k-1) are respectively the spheroidization matrix calculated based on the data obtained at time k and the spheroidization matrix calculated based on the data obtained at time k-1.

S24: Convert the SFA transformation matrix W and the normalized matrix Multiply to obtain slow features s.

The oscillation signal caused by valve stiction has a large autocorrelation, so possible oscillations can be extracted in the form of slow feature information. In addition, some noise has zero lag autocorrelation, so a large degree of noise removal can be achieved in this process.

S3: Calculate the Hurst index of the slowest feature among the slow features.

The SFA method only extracts single-lag temporal dynamic feature information. In order to further extract long-term correlation information, the Hurst index is used to analyze the extracted slowest feature s ₁ to obtain the long-term correlation of the slowest feature for detection and analysis. Nonlinear behavior in the system.

The slow features extracted by S3 are arranged in order from slow to fast change. Among them, the slow feature s ₁ that changes the slowest is the slowest feature.

The Detrended Fluctuation Analysis (DFA) method is used to calculate the Hurst index of the slowest feature among the slow features, and the long-term correlation of the slowest feature is obtained, which is used to quantify nonlinear characteristics. The process of calculating the Hurst index includes:

Calculate the cumulative difference of the slowest feature and obtain the cumulative difference sequence;

Divide the cumulative difference sequence into multiple windows, and for each division, calculate the root mean square value of the sequence under different windows, and calculate the logarithm of the window length and the logarithm of the root mean square value of the sequence under different windows;

Perform linear fitting on the two logarithmic values and obtain the slope of the linear fitting, which is the Hurst index.

Use formula (9) to calculate the cumulative difference of the slowest feature s ₁ , and obtain the cumulative difference sequence S(i).

For each division, the cumulative difference sequence is split into q non-overlapping windows of length n, and then a first-order least squares line is fitted to each window And calculate the variance under each window/>

Calculate the root mean square value F(n) of the sequence under different windows:

After that, log[F(n)] and log(n) are calculated.

The value of the window length n is different for each division.

Perform linear fitting on the calculated log[F(n)] and log(n) to obtain the slope of the linear fitting, which is the Hurst index He.

S4: Determine the valve static friction detection index based on the Hurst index.

According to the Hurst index and the static friction detection index model, the valve static friction detection index is determined, where the static friction detection index model is:

In the formula, He is the Hurst index; Hs is the valve static friction detection index.

The Hs value ranges from 0 to 1. The closer the value is to 0, the higher the long-term correlation in the time series, that is, the presence of strong nonlinear characteristics in the time series data, which is a sign of valve stickiness.

S5: Determine whether the valve has static friction failure according to the valve static friction detection index, and obtain the fault detection result.

When the valve static friction detection index is less than the set threshold, it is determined that the valve has a static friction failure.

Preferably, the threshold is set to 0.5.

If Hs is less than 0.5, it indicates valve sticking failure.

The method disclosed in this embodiment collects and preprocesses OP and PV data under valve static friction failure, uses SFA as the first layer of machine learning to extract slowly changing time series dynamic features in the data, and performs the extraction on the slowest features after extraction. The Hurst index of the second layer calculates the long-term correlation information in s ₁ , and finally detects the static friction of the valve. This method improves the anti-interference ability against noise and extracts the timing dynamic characteristics and long-term correlation information of the entire timing data process. , which improves the valve static friction detection performance, especially the ability to accurately detect the process non-periodic random oscillation caused by valve static friction.

The method disclosed in this embodiment extracts the time series dynamic characteristic information in the OP-PV reconstructed data and eliminates the noise signal at the same time, which improves the robustness of the detection method to the noise signal; according to the nonlinear characteristics of the valve static friction, the process data will be long. The time correlation is significantly increased, taking full account of the significant long-term correlation of process data, improving the valve static friction detection accuracy, especially improving the detection performance when static friction causes irregular oscillations in the control loop.

Example 2

In this embodiment, a process control system valve stiction fault detection system is disclosed, including:

Data acquisition module, used to obtain the controller output data and controlled process variable data of the valve;

Slow feature extraction module, used to extract slow features from controller output data and controlled process variable data;

The Hurst index calculation module is used to calculate the Hurst index of the slowest feature among the slow features;

The valve static friction detection index acquisition module is used to determine the valve static friction detection index based on the Hurst index;

The fault detection result acquisition module is used to determine whether a static friction fault occurs in the valve according to the valve static friction detection index and obtain the fault detection result.

Example 3

In this embodiment, an electronic device is disclosed, including a memory, a processor, and computer instructions stored in the memory and executed on the processor. When the computer instructions are executed by the processor, a method disclosed in Embodiment 1 is completed. The steps described in a process control system valve stiction fault detection method.

Example 4

In this embodiment, a computer-readable storage medium is disclosed for storing computer instructions. When the computer instructions are executed by a processor, the method for detecting stiction failure of valves in a process control system disclosed in Embodiment 1 is completed. A step of.

Finally, it should be noted that the above embodiments are only used to illustrate the technical solution of the present invention and not to limit it. Although the present invention has been described in detail with reference to the above embodiments, those of ordinary skill in the art should understand that the present invention can still be modified. Modifications or equivalent substitutions may be made to the specific embodiments, and any modifications or equivalent substitutions that do not depart from the spirit and scope of the invention shall be covered by the scope of the claims of the invention.

Claims (10)

1. A process control system valve static friction fault detection method, which is characterized by including: Obtain controller output data and controlled process variable data of the valve; Extract slow features from controller output data and controlled process variable data; Calculate the Hurst index of the slowest feature among the slow features; According to the Hurst index, determine the valve static friction detection index; According to the valve static friction detection index, it is judged whether the valve has static friction failure and the fault detection result is obtained. 2. A process control system valve static friction fault detection method as claimed in claim 1, characterized in that the controller output data and the controlled process variable data are preprocessed to obtain the preprocessed data; Extract slow features from preprocessed data. 3. A process control system valve static friction fault detection method according to claim 2, characterized in that the preprocessing process includes standardization processing, data reconstruction and dimension expansion. 4. A process control system valve static friction fault detection method according to claim 2, characterized in that a slow feature analysis algorithm is used to extract slow features from preprocessed data. 5. A process control system valve static friction fault detection method according to claim 1, characterized in that the DFA method is used to calculate the Hurst index of the slowest feature among the slow features. 6. A process control system valve static friction fault detection method as claimed in claim 1, characterized in that the valve static friction detection index is determined according to the Hurst index and the static friction detection index model; wherein the static friction detection index model is: In the formula, He is the Hurst index; Hs is the valve static friction detection index. 7. A process control system valve static friction fault detection method according to claim 1, characterized in that when the valve static friction detection index is less than the set threshold, it is determined that the valve has a static friction fault. 8. A process control system valve static friction fault detection system, which is characterized by including: Data acquisition module, used to obtain the controller output data and controlled process variable data of the valve; Slow feature extraction module, used to extract slow features from controller output data and controlled process variable data; The Hurst index calculation module is used to calculate the Hurst index of the slowest feature among the slow features; The valve static friction detection index acquisition module is used to determine the valve static friction detection index based on the Hurst index; The fault detection result acquisition module is used to determine whether a static friction fault occurs in the valve according to the valve static friction detection index and obtain the fault detection result. 9. An electronic device, characterized in that it includes a memory and a processor and computer instructions stored in the memory and run on the processor. When the computer instructions are run by the processor, any one of claims 1-7 is completed. The steps of the method for detecting static friction failure of valves in a process control system. 10. A computer-readable storage medium, characterized in that it is used to store computer instructions. When the computer instructions are executed by a processor, the process control system valve stiction failure described in any one of claims 1-7 is completed. Steps of the detection method.