CN104865952A - Method for setting double-ellipse fitting through grid frequencies - Google Patents

Abstract

The invention discloses a method for setting double-ellipse fitting through grid frequencies. The method comprises the following steps of: selecting cut-off frequency groups composed of upper cut-off frequencies and lower cut-off frequencies in upper cut-off frequencies and lower cut-off frequencies; calculating the percentage of the count of a subset data segment between an inner ellipse and an outer ellipse in all counts of a whole subset data segment according to each cut-off frequency group; and finally obtaining the maximum percentage and the upper cut-off frequency and the lower cut-off frequency corresponding to the maximum percentage. The selection process of the upper cut-off frequencies and the lower cut-off frequencies of a band-pass filter is optimized. According to the method, the detection rate of viscosity property can be increased, and the degree of automation of algorithms is increased.

Description

Grid Frequency Setting Double Ellipse Fitting Method

This application is a divisional application with an application date of May 02, 2013, an application number of 201310159959.5, and an invention title of "Grid frequency setting double ellipse fitting method and its application".

technical field

The invention belongs to the technical field of a control circuit performance monitoring system, and in particular relates to a method for detecting a sticking fault in a pneumatic control valve.

Background technique

In continuous process industries such as electric power, steel, petroleum, chemical industry, etc., with the increasing production scale, the installed capacity of a single set of equipment is getting larger and larger, and it is very important for energy saving, cost control, production safety, product quality, and environmental protection. The requirements for indicators such as are also constantly improving, resulting in the role of process control in the entire production becoming more and more important. However, due to various reasons such as the increasing complexity of modern industrial processes and the lack of daily maintenance of a large number of controllers on site, there are a large number of problems of poor performance of control loop systems in industrial processes. According to a 2-year investigation on the continuous process industrial process released by Honeywell, the performance of 26,000 PID control loops in the continuous process industrial process was analyzed, and the results showed that only 1/3 of the control loop systems performed well. And the system performance of the other 2/3 control loops needs to be improved. A major problem that makes control loops perform poorly is loop oscillations. A survey shows that 1/3 of the control loops in the industrial process are in an oscillating state. The existence of oscillation directly leads to problems such as product quality decline, energy consumption increase, and equipment wear and tear. Especially for large-scale continuous process industries, the number of control loops is tens of thousands, and it is particularly urgent to eliminate the adverse effects of loop oscillation on production.

There are three main reasons for loop oscillation, namely: process equipment problems, periodic external disturbances, and poor parameter setting of the controller. Control valve is one of the most commonly used process equipment in industrial processes such as electric power, steel, petroleum, and chemical industry. Among the various reasons that lead to loop oscillation, the problem of the pneumatic actuator valve accounts for 20% to 30% according to statistics. Non-linear characteristics such as hysteresis, dead zone, and viscosity of the pneumatic actuator valve will affect the performance of the control system to varying degrees. , and among them, the viscous characteristics of pneumatically actuated valves are the most common. According to the literature, a 1% improvement in control system performance or energy utilization in industrial processes will bring tens of millions or even hundreds of millions of dollars in profit. Therefore, it is necessary to strengthen the online real-time monitoring of the production process and the operation status of the control system, realize the online diagnosis of various actuator faults including the oscillation problem caused by valve sticking, and take corresponding solutions accordingly. The safety, stability and efficiency of continuous process production such as electric power, steel, petroleum and chemical industry are very important.

The method proposed in the existing literature to detect the stickiness of the pneumatic control valve is to use the bicoherent method to detect the nonlinearity of the control loop, and confirm the existence of the sticky characteristic of the pneumatic control valve by fitting the ellipse, and give a measure of the stickiness of the pneumatic control valve. estimate. However, due to the complex factors such as a large amount of noise in the actual field data used for the above analysis and calculation, the interference of the "ellipse fitting" test results, and some cases of actual stickiness, if the methods in the existing literature are used, However, an estimate of the presence of stickiness cannot be obtained. That is to say, the methods for detecting the stickiness of the pneumatic control valve proposed in the existing literature have the problem that the detection effectiveness is not ideal.

Contents of the invention

Purpose of the invention: The present invention provides a grid frequency setting double ellipse fitting method, which can improve the detection rate of the stickiness of the pneumatic control valve, thereby further improving the degree of automation of the algorithm.

Technical solution: a grid frequency setting double ellipse fitting method, wherein the selection process of the upper cutoff frequency and the lower cutoff frequency of the bandpass filter is as follows:

S1, setting the upper limit and lower limit of the upper cutoff frequency of the bandpass filter, the upper limit and lower limit of the lower cutoff frequency, and the upper cutoff frequency step size and the lower cutoff frequency step size;

S2. Search the upper cut-off frequency and the lower cut-off frequency, and calculate the percentage of the number of points in the subset data segment between the inner ellipse and the outer ellipse to all points in the entire subset data segment;

S2.11. For a given upper cut-off frequency, the lower cut-off frequency decreases downward from the upper limit of the lower cut-off frequency or increases upward from the lower limit of the lower cut-off frequency according to the lower cut-off frequency step size. The upper cut-off frequency and the lower cut-off frequency of , calculate the percentage of the subset data segment points between the inner ellipse and the outer ellipse to all the points of the entire subset data segment;

S2.12. Between the upper limit and the lower limit of the upper cut-off frequency, change the value of the upper cut-off frequency according to the step size of the upper cut-off frequency, and repeat step S2.11;

Get and output the maximum percentage among the percentages, and record the upper cut-off frequency and lower cut-off frequency corresponding to this percentage;

or;

S2.21. For a given lower cut-off frequency, the upper cut-off frequency decreases downward from the upper limit of the upper cut-off frequency or increases upward from the lower limit of the upper cut-off frequency according to the step size of the upper cut-off frequency. The upper cut-off frequency and the lower cut-off frequency of , calculate the percentage of the subset data segment points between the inner ellipse and the outer ellipse to all the points of the entire subset data segment;

S2.22. Between the upper limit and the lower limit of the lower cut-off frequency, change the value of the lower cut-off frequency according to the step size of the lower cut-off frequency, and repeat step S2.21;

S2.3. Obtain and output the maximum percentage among the percentages, and record the upper cut-off frequency and lower cut-off frequency corresponding to this percentage;

S3. If the maximum percentage is greater than or equal to the percentage threshold, it is judged that sticking occurs.

In step S2, the method of searching for the largest percentage among the percentages is: comparing the percentages obtained later with the percentages obtained earlier, if the percentage obtained later is greater than the percentage obtained earlier, the percentage obtained later is kept and recorded The upper and lower cutoff frequencies; otherwise, the percentage obtained after discarding.

In step S2, the method of searching for the maximum percentage in the percentage is: record the number of points in the subset data segment between the inner ellipse and the outer ellipse corresponding to each group of upper limit frequency and lower limit frequency to account for the percentage of all points in the entire subset data segment percentage, and pick the largest percentage out of them.

The inner ellipse is obtained by reducing the standard ellipse by the first scaling factor; the outer ellipse is obtained by enlarging the standard ellipse by the second scaling factor. The first scaling factor is equal to the second scaling factor. The first scaling factor and the second scaling factor are 12%-18%, preferably, the first scaling factor and the second scaling factor are 15%.

In the above method, the percentage threshold is 60%; the upper limit of the upper cut-off frequency of the bandpass filter is 0.5, and the lower limit is 0.02; the upper limit of the lower cut-off frequency is less than 0.02, and the lower limit is 0.001; the upper cut-off frequency step The length is 0.01~0.05, and the lower cutoff frequency step is 0.001~0.002.

In the above S2, the method of searching for the upper cut-off frequency and the lower cut-off frequency can also be:

The upper cut-off frequency decreases downward from the upper limit of the upper cut-off frequency according to the upper cut-off frequency step, or increases upward from the lower limit of the upper cut-off frequency to obtain multiple sets of given upper cut-off frequencies; the lower cut-off frequency increases from the lower cut-off frequency to The upper limit of the cut-off frequency decreases downward, or increases upward from the lower limit of the lower cut-off frequency to obtain multiple sets of given lower cut-off frequencies; any given upper cut-off frequency and any given lower cut-off frequency form a cut-off frequency group , for each cutoff frequency group, calculate the percentage of the number of points in the subset data segment between the inner ellipse and the outer ellipse to all points in the entire subset data segment, and obtain the maximum percentage and the corresponding upper and lower cutoff frequencies Cut-off frequency.

The present invention also provides the application of the above-mentioned grid frequency setting double ellipse fitting method in the control valve viscosity detection process.

Beneficial effects: the grid frequency setting double ellipse fitting method of the present invention can increase the detection rate of the viscous effect, thereby improving the degree of automation of the algorithm and obtaining considerable economic benefits.

Description of drawings

Fig. 1 is a flow chart of the present invention;

Fig. 2a～Fig. 2c are the experimental data figures of an embodiment of the present invention;

3a to 3c are diagrams of experimental data of another embodiment of the present invention.

Detailed ways

Example 1:

As shown in Figure 1, the basic steps of the grid frequency setting double ellipse fitting method include:

S1, assignment, initialization, setting the upper limit f _Hmax and lower limit f _Hmin of the upper cut-off frequency of the bandpass filter, the upper limit f _Lmax and the lower limit f _Lmin of the lower cut-off frequency, and the upper cut-off frequency step Step_H and the lower cut-off frequency step Step_L, assign 0 to i and P _max ; for example, the upper limit f _Hmax of the upper cut-off frequency of the bandpass filter is 0.5, and the lower limit f _Hmin is 0.02; the upper limit f _Lmax of the lower cut-off frequency is less than 0.02, and the lower limit f _Lmin is 0.001; The step size of the upper cut-off frequency Step_H is 0.01-0.05, and the step size of the lower cut-off frequency Step_L is 0.001-0.002.

S2. Use the double loop method to search for the upper and lower cut-off frequencies, the lower cut-off frequency is the outer loop, and the upper cut-off frequency is the inner loop. If the value of a certain lower cut-off frequency f _Lmax -i*Step_L is less than f _Lmin , it is considered that the lower cut-off frequency has been traversed from top to bottom. At this time, the inner loop has also been traversed, then output P _max , if the value is greater than If the percentage threshold is set, stickiness is considered to exist, otherwise, stickiness is not considered to exist.

In the inner loop, the upper cut-off frequency traverses from top to bottom. If the value of a certain upper cut-off frequency f _Hmax -j*Step_H is less than f _Hmin , it is considered that an inner loop is completed; otherwise, the calculation is performed in {f _{L( i)} , f _H(j) } take the value of P(i, j), record it as P _new , add 1 to j, and prepare to enter the next cycle; judge whether P _new is greater than P _max , if P _new is greater than P _max, Then assign P _new to P _max so that P _max is always the maximum value of the calculated percentage value; otherwise, re-enter the next step of the inner loop to recalculate a set of percentages.

Example 2:

S1, assignment, initialization, setting the upper limit f _Hmax and lower limit f _Hmin of the upper cut-off frequency of the bandpass filter, the upper limit f _Lmax and the lower limit f _Lmin of the lower cut-off frequency, and the upper cut-off frequency step Step_H and the lower cut-off frequency step Step_L, assign 0 to i and P _max ; for example, the upper limit f _Hmax of the upper cut-off frequency of the bandpass filter is 0.5, and the lower limit f _Hmin is 0.02; the upper limit f _Lmax of the lower cut-off frequency is less than 0.02, and the lower limit f _Lmin is 0.001; The step size of the upper cut-off frequency Step_H is 0.01-0.05, and the step size of the lower cut-off frequency Step_L is 0.001-0.002. S2. Search the upper cut-off frequency and the lower cut-off frequency, and calculate the percentage of the subset data segment points between the inner ellipse and the outer ellipse accounting for all the points of the entire subset data segment; the inner ellipse is reduced by the first scaling factor by the standard ellipse Obtain; the outer ellipse is obtained by enlarging the standard ellipse by the second scaling factor, the first scaling factor is equal to the second scaling factor, the first scaling factor and the second scaling factor are generally 12% to 18%, preferably 15%.

Step S2 specifically includes: S2.11. For a given upper cut-off frequency f _H , the lower cut-off frequency f _L decreases from the upper limit f _Lmax downwards or increases upwards from the lower limit f _Lmin according to the lower cut-off frequency step Step_L , for each given upper cut-off frequency and lower cut-off frequency, calculate the percentage P _new of the subset data segment points between the inner ellipse and the outer ellipse to all the points of the entire subset data segment; the percentage obtained after comparison is with The size of the percentage obtained before, if the percentage obtained later is greater than the percentage obtained first, keep the percentage obtained later, and record the upper cut-off frequency and lower cut-off frequency; otherwise, discard the percentage obtained after.

S2.12. Between the upper limit f _Hmax and the lower limit f _Hmin of the upper cut-off frequency, change the value of the upper cut-off frequency f _H according to the step size of the upper cut-off frequency Step_H, and repeat step S2.11;

Obtain the maximum percentage P _max in the percentage P _new , and record the upper cut-off frequency and lower cut-off frequency corresponding to this percentage;

S3. If the maximum percentage P _max is greater than or equal to the percentage threshold P _set , it is determined that sticking occurs. According to engineering experience, the percentage threshold P _set is 60%.

Example 3:

The difference between this embodiment and the previous implementation is in step S2:

S2.21. For a given lower cut-off frequency f _L , the upper cut-off frequency f _H decreases downward from the upper limit f _Hmax or increases upward from the lower limit f _Hmin according to the upper cut-off frequency step Step_H, and each group is given Determine the upper cut-off frequency and the lower cut-off frequency, and calculate the percentage P _new of the number of points in the subset data segment between the inner ellipse and the outer ellipse to all points in the entire subset data segment;

S2.22. Between the upper limit f _Lmax and the lower limit f _Lmin of the lower cut-off frequency, change the value of the lower cut-off frequency f _L according to the lower cut-off frequency step Step_L, and repeat step S2.21; obtain the maximum percentage P _max in the percentage P _new , record the upper and lower cutoff frequencies corresponding to this percentage.

Example 4:

The difference between this embodiment and embodiment 2 is that: in step S2, the method of searching for the maximum percentage P _max in the percentage P _new is: recording the inner ellipse and the outer ellipse corresponding to each group of upper limit frequency and lower limit frequency The percentage of the number of points in the subset data segment among all the points in the entire subset data segment, and select the largest percentage from it.

Example 5:

The difference between this embodiment and Embodiment 2 is that: in said S2, the method for searching the upper cut-off frequency and the lower cut-off frequency can also be: the upper cut-off frequency is from the upper limit of the upper cut-off frequency downwards according to the upper cut-off frequency step size Decrease or increase upward from the lower limit of the upper cutoff frequency to obtain multiple sets of given upper cutoff frequencies; the lower cutoff frequency decreases from the upper limit of the lower cutoff frequency according to the lower cutoff frequency step, or increases upward from the lower limit of the lower cutoff frequency , to obtain multiple sets of given lower cutoff frequencies; combine any given upper cutoff frequency and any given lower cutoff frequency to form a cutoff frequency group, and for each cutoff frequency group, calculate between the inner ellipse and the outer ellipse The percentage of points in the subset data segment accounted for the percentage of all points in the entire subset data segment, and obtain the maximum percentage and the upper cut-off frequency and lower cut-off frequency corresponding to the maximum percentage.

From the above embodiments, it can be seen that no matter whether the upper cutoff frequency is the inner loop, the lower cutoff frequency is the outer loop, or vice versa, the traversal can be realized, and the upper and lower cutoff frequencies can be traversed from top to bottom or from bottom to top. In addition, the methods for finding the maximum percentage are not limited to the above-mentioned ones, and technicians can find other algorithms from existing algorithms.

Comparative experiment 1: As shown in Figure 2a to Figure 2c, this is the OP (control signal output to the actuator), PV (actual value detected) drawn according to the industrial data of a viscous flow control loop. ), SP (setting value) and PV-OP diagram. Filter the data (OP and PV) according to the upper and lower cut-off frequency parameters of the optimized band-pass filter determined by the present invention, and then fit them according to the double ellipse to obtain =62%. Since =62%>=60%, it is judged that the ellipse fitting is successful, and the fitting result AS=0.18 when optimizing the frequency, it can be judged that there is stickiness in the loop, which is completely consistent with the actual situation. However, according to the selection method of the upper and lower cut-off frequencies of the bandpass filter in the prior art, after filtering, the double ellipse is used to fit, and =41.8% is obtained. Since =41.8%<=60%, it is judged that the ellipse fitting fails, and the wrong result is obtained, that is, the loop is not viscous.

Comparative Experiment 2: As shown in Figure 3a to Figure 3c, this is the OP, PV, SP and PV-OP diagrams drawn according to the industrial data of a liquid level control with actual viscosity. Filter the data (OP and PV) according to the upper and lower cut-off frequency parameters of the optimized band-pass filter determined by the present invention, and then fit them according to the double ellipse to obtain =98%. Since =98%>=60%, it is judged that the ellipse fitting is successful, and the fitting result AS=1.18 when optimizing the frequency, so it can be judged that there is stickiness in the loop, which is completely consistent with the actual situation. However, according to the prior art method for selecting the upper and lower cutoff frequencies of the bandpass filter, after filtering, the double ellipse is used to fit, and =41.7% is obtained. Since =41.7%<=60%, it is judged that the ellipse fitting fails, and the wrong result is obtained, that is, the loop is not viscous.

In the above experiment, the value of Pset is 60%, but those skilled in the art can choose different values according to different situations.

Claims (8)

1. a grid frequency setting double ellipse fitting method is characterized in that the selection process of the upper cut-off frequency and the lower cut-off frequency of the bandpass filter is as follows: S1. Set the upper limit (f _Hmax ) and lower limit (f _Hmin ) of the upper cut-off frequency of the band-pass filter, the upper limit (f _Lmax ) and lower limit (f _Lmin ) of the lower cut-off frequency, and the step size of the upper cut-off frequency (Step_H) and the lower cutoff frequency step (Step_L); S2. Search the upper cut-off frequency and the lower cut-off frequency, and calculate the percentage of the number of points in the subset data segment between the inner ellipse and the outer ellipse to all points in the entire subset data segment; S2.11. For a given upper cut-off frequency (f _H ), the lower cut-off frequency (f _L ) increases upwards from the lower limit (f _Lmin ) of the lower cut-off frequency according to the lower cut-off frequency step (Step_L), for each Given the upper cut-off frequency and lower cut-off frequency of the group, calculate the percentage of the subset data segment points between the inner ellipse and the outer ellipse to all the points of the entire subset data segment ( P _new ); S2.12. Between the upper limit of the upper cut-off frequency (f _Hmax ) and the lower limit of the upper cut-off frequency (f _Hmin ), change the value of the upper cut-off frequency (f _H ) according to the upper cut-off frequency step (Step_H), and repeat step S2 .11; Obtain and output the maximum percentage ( P _max ) of the percentages ( P _new ), record the upper cut-off frequency and lower cut-off frequency corresponding to this percentage; S3. If the maximum percentage ( P _max ) is greater than or equal to the percentage threshold ( P _set ), it is determined that sticking occurs. 2. grid frequency setting double ellipse fitting method as claimed in claim 1 is characterized in that, in step S2, The method of searching for the maximum percentage ( P _max ) in the percentage ( P _new ) is: compare the size of the percentage obtained later with the percentage obtained before, if the percentage obtained later is greater than the percentage obtained first, then keep the percentage obtained later , and record the upper cutoff frequency and lower cutoff frequency; otherwise, discard the obtained percentage. 3. The grid frequency setting double ellipse fitting method according to claim 1, characterized in that, in step S2, the method of searching for the maximum percentage ( P _max ) among the percentages ( P _new ) is: record each The percentage of points in the subset data segment between the inner ellipse and the outer ellipse corresponding to the upper limit frequency and the lower limit frequency of the group account for all the points in the entire subset data segment, and select the maximum percentage from it. 4. grid frequency setting double ellipse fitting method as claimed in claim 1 or 2, is characterized in that, described inner ellipse is obtained by shrinking the first scaling factor by standard ellipse; Described outer ellipse is obtained by standard ellipse by the first Two scaling factors are obtained by zooming in. 5. The grid frequency setting double ellipse fitting method according to claim 4, characterized in that, the first scaling factor is equal to the second scaling factor. 6. The grid frequency setting double ellipse fitting method according to claim 5, wherein the first scaling factor and the second scaling factor are 12% to 18%. 7. The grid frequency setting double ellipse fitting method according to claim 6, wherein the first scaling factor and the second scaling factor are 15%. 8. The grid frequency setting double ellipse fitting method as claimed in claim 1, characterized in that, the percentage threshold ( P _set ) is 50-70%; the upper limit of the upper cut-off frequency of the band-pass filter (f _Hmax ) is 0.5, the lower limit (f _Hmin ) is 0.02; the upper limit (f _Lmax ) of the lower cutoff frequency is less than 0.02, and the lower limit (f _Lmin ) is 0.001; the upper cutoff frequency step (Step_H) is 0.01~0.05, The lower cutoff frequency step size (Step_L) is 0.001~0.002.