Two-out-of-three control method for main connection displacement sensor based on correlation

Abstract

The invention provides a correlation-based main connection displacement sensor two-out-of-three control method, which comprises the following steps of: 1, acquiring code values # imgabs0 #, # imgabs1 # and # imgabs2 # output by three sensors 1 #, 2 # and 3 # in real time, and respectively calculating mean values # imgabs3 #, # imgabs4 # and # imgabs5 # of n acquired data; step2, correlation coefficients among the three sensors are calculated, deviation values of all the correlation coefficients and complete positive correlation in the current sampling period are calculated respectively, and preliminary judgment is carried out, Step3, the Step2 is repeated to calculate all the correlation coefficients # imgabs6 #, # imgabs7 # and # imgabs8 # in the next sampling period, deviation values of all the correlation coefficients and complete positive correlation # imgabs9 #, # imgabs10 # and # imgabs11 # are judged, and the deviation values of all the correlation coefficients and complete positive correlation are calculated. And a fault sensor is judged by using a two-out-of-three logic, and adopted sensor data is selected. Two-out-of-three logic judgment of sensor fault judgment can be carried out by calculating the Pearson correlation coefficient between sensor output values and the unit state, and the method can be achieved without alignment installation of sensors of the same type or data per unit.

Description

Technical Field

[0001] The present invention relates to the technical field of hydro-generator control, and in particular to a correlation-based two-out-of-three control method for main-connected displacement sensors. Background Art

[0002] The main servomotor displacement sensor measures the linear displacement of the main servomotor. The measured data is used for closed-loop control of the speed control system and is a critical component of the turbine speed control system. Its data quality is directly related to the stability of the speed control system's dynamic control. Medium- and large-scale hydropower stations generally adopt a three-out-of-two configuration: three sensors of the same type are installed next to the servomotor. The movable portion of the sensor is connected to the servomotor piston rod and moves linearly with the piston rod. Specifically, the three sensors are arranged in parallel next to the servomotor, with a transverse bracket installed perpendicular to the junction of the servomotor piston rod and the control loop. The movable portions of the three sensors are connected to the transverse bracket and move with the servomotor. The sensors output a 4-20mA current to controllers A and B. The controllers convert the current value into a code value, which is then used for fault diagnosis and three-out-of-two master selection. Sensor 1 sends the data directly to controller A and then to controller B via dual-machine communication. Sensor 2 sends the data directly to sensor B and then to controller A via dual-machine communication. Sensor 3 sends the data to controllers A and B via a split-to-two transmitter.

[0003] The existing method for selecting two-out-of-three configurations of the three main relay displacement sensors is to first determine sensor disconnection and jump faults, and then perform two-out-of-three master selection.

[0004] Disconnection fault judgment: When the sensor sampling code value is less than the set lower limit or greater than the set upper limit for 5 sampling cycles, it is judged as a disconnection fault.

[0005] Jump fault judgment: The controller calculates the absolute value of the difference between the current and future sampling period sampling values ​​and the sampling value of the previous sampling period. If the difference of 5 consecutive sampling periods is greater than the multiple of the set alarm threshold, it is judged as a jump fault. Taking sensor 1# as an example, .in It is the sampling value of sensor 1# in the previous sampling period; is the sampling value of the current sampling period, is the sampling value of the next sampling period, is the sampling value of the 5th sampling cycle; The set alarm threshold.

[0006] Two-out-of-three master selection: When all three sensors are normal, the difference between the sensors is calculated. When the absolute values ​​of the deviations are all less than the alarm threshold, the master sensor is used for control. When the absolute value of any deviation is greater than the alarm threshold, a minor deviation fault alarm is issued, and control is still carried out using the master sensor. When the absolute values ​​of both deviations are greater than the alarm threshold, a major deviation fault alarm is issued, and the standby machine is switched to operation.

[0007] The aforementioned two-out-of-three approach has the following drawbacks: 1. When a sensor experiences a slow and persistent linearity shift, existing jump fault detection methods cannot detect the fault based on its own data. 2. The two-out-of-three judgment logic requires that sensors be aligned and of the same type, or that the data be per-unit processed. This is particularly complex and cumbersome when disassembling and installing sensors during overhauls. Summary of the Invention

[0008] The technical problem to be solved by the present invention is to provide a correlation-based three-to-two control method for main displacement sensors, which solves the problems that the existing three-to-two method for sensor data cannot deal with the slow linear offset of the sensor and must be of the same type and have data standardization to be implemented.

[0009] In order to solve the above technical problems, the technical solution adopted by the present invention is: A correlation-based two-out-of-three control method for main displacement sensors, the method comprising: Step 1: Real-time acquisition of the code values ​​output by three sensors 1#, 2#, and 3# , , , i represents the data collected in real time according to the collection cycle, and takes values ​​1, 2, 3...n; and calculates n The mean of the collected data 、 、 ;Step 2, calculate the distance between the three sensors n Correlation coefficient of sampling period 、 、 ; Calculate the deviation value of each correlation coefficient and the perfect positive correlation in the current sampling period respectively 、 and ;like ,and and , then there is a strong correlation between 1# and 2# sensors, and the abnormal count of 3# sensor increases by 1. This is the abnormality judgment threshold, which is adjusted according to the accuracy of the sensor; Step 3: Repeat Step 2 to calculate the correlation coefficients in the next sampling period. 、 、 , and determine the deviation value of each correlation coefficient from the complete positive correlation 、 and , and uses three-choose-two logic to determine the faulty sensor and select the sensor data to be adopted.

[0010] In Step 1 above n The mean of the data in the sampling period 、 、 The calculation formula is: ; ; ; in n Set according to the sampling period.

[0011] The calculation formula for the correlation coefficient between the three sensors in Step 2 above is: ; ; .

[0012] The deviation value of each correlation coefficient from the completely positive correlation within the sampling period in Step 2 above 、 and The calculation formula is: ; ; .

[0013] The specific method for the three-choose-two logic judgment in Step 3 above is: like ,and and , then the 3# sensor abnormality count is increased by 1, otherwise the 3# sensor abnormality count is set to 0; if the 3# sensor abnormality count value is equal to 5 n , then the output is 3# sensor fault; Similarly, if ,and and , then the abnormal count of sensor 2# is increased by 1, otherwise the abnormal count of sensor 2# is set to 0; if the abnormal count value of sensor 2# is equal to 5 n , then the output is 3# sensor fault; like ,and and , then the abnormal count of sensor 1# is increased by 1, otherwise the abnormal count of sensor 1# is set to 0; if the abnormal count value of sensor 1# is equal to 5 n , then the output 3# sensor failure.

[0014] In the above Step 3, if 、 、 Are less than or equal to , then 1#, 2#, 3# sensors are normal and have no faults; if 、 、 Both greater than , and the difference between the guide vane setting and the guide vane feedback is greater than the adjustment dead zone, the abnormal counts of 1#, 2#, and 3# sensors are all set to 1. When the abnormal count values ​​of the sensors are all equal to 5 n , then the output 1#, 2#, 3# sensor failure.

[0015] In the above Step 3 three-choose-two logic judgment, when 5 consecutive n Sampling period 、 、 When both are established and there is no deviation fault, the current main operation is maintained; 5 in a row n Sampling period 、 、 If all the above conditions are met, it is determined that the 1# sensor has a deviation fault, and the 2# sensor of the B set controller is switched to the main operation; 5 in a row n Sampling period 、 、 If all the above conditions are met, it is determined that the 2# sensor has a deviation fault, and the 1# sensor of the A set of controllers is switched to the main operation; 5 in a row n Sampling period 、 、 If all of the above are true, it is determined that the 3# sensor has a deviation fault and the main function is maintained; 5 in a row n sampling period and the difference between the guide vane setting and the guide vane feedback is greater than the regulation dead zone, 、 、 If both are true, it is determined that the sensor is abnormal. At this time, it is necessary to use historical correlation data to determine which sensor is working properly and use it as the main sensor. T The total samples in time are N Historical data 、 、 Less than or equal to The number of samples N AB 、N AC 、N BC , calculate the correlation frequency between each pair of sensors 、 、 ; Then calculate the reliability frequency of a single sensor 、 、 ; Define sensor reliability weight W , for sensor A, , calculate the weighted deviation of the difference between sensor A and the other two sensors , calculate the comprehensive failure probability of sensor A Similarly, calculate the comprehensive failure probability of sensors B and C P B 、P C ,Comparing the sizes of the three, the smallest one is used as the main sensor, and the other two are marked as abnormal.

[0016] In the above Step 3, the two-choose-three logic judgment and the master selection judgment are carried out according to the following table: Three-choice-two primary selection decision table .

[0017] The movable parts of the above three sensors 1#, 2# and 3# are all connected to the transverse bracket and move with the relay. The transverse bracket is installed in the vertical direction of the connection between the relay piston rod and the control ring. The three sensors are arranged in parallel next to the relay.

[0018] The correlation-based two-out-of-three control method for main displacement sensors mentioned in the present invention has the following beneficial effects: 1. This three-choose-two control method is suitable for judgment of sensors of different types, and the control is simple and easy to implement.

[0019] 2. This three-choose-two control method does not require sensor alignment installation. When the unit is overhauled and disassembled to install the sensor, there is no need to mark the position, which improves work efficiency.

[0020] 3. This three-choose-two control method does not require per-unit processing of the data. The detection accuracy can be freely adjusted by adjusting the parameters according to the actual situation on site. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] The present invention will be further described below with reference to the accompanying drawings and examples: Figure 1 This is a heat map of correlation coefficients between the sampling values ​​of three sensors in an embodiment of the present invention; Figure 2 The sampling value curve when the sensor is fault-free in the embodiment of the present invention; Figure 3 The correlation coefficient curve of the sampling values ​​when the sensor is fault-free in the embodiment of the present invention; Figure 4 This is the dead value fault sampling value curve of sensor 3# in the embodiment of the present invention; Figure 5 This is the correlation coefficient curve of the dead value fault sampling value of sensor 3# in the embodiment of the present invention; Figure 6 This is the sampling value curve of the 3# sensor jump fault in the embodiment of the present invention; Figure 7 This is the correlation coefficient curve of the 3# sensor jump fault sampling value in the embodiment of the present invention; Figure 8 This is the judgment process when two sensors are normal and one sensor is faulty in the present invention; Figure 9 This is the judgment process when all three sensors fail in the present invention. DETAILED DESCRIPTION

[0022] The technical solution of the present invention is described in detail below with reference to the accompanying drawings and embodiments.

[0023] Example 1: A correlation-based two-out-of-three control method for main displacement sensors includes: Step 1: Real-time acquisition of the code values ​​output by three sensors 1#, 2#, and 3# , , , i represents the data collected in real time according to the collection cycle, and takes values ​​1, 2, 3...n; and calculates n The mean of the collected data 、 、 ;Step 2, calculate the distance between the three sensors n Correlation coefficient of sampling period 、 、 ; Calculate the deviation value of each correlation coefficient and the perfect positive correlation in the current sampling period respectively 、 and ;like ,and and , then there is a strong correlation between 1# and 2# sensors, and the abnormal count of 3# sensor increases by 1. The abnormality judgment threshold is set according to the accuracy of the sensor, and is generally set to 0.001-0.05; Step 3: Repeat Step 2 to calculate the correlation coefficients in the next sampling period. 、 、 , and determine the deviation value of each correlation coefficient from the complete positive correlation 、 and , and uses three-choose-two logic to determine the faulty sensor and select the sensor data to be adopted.

[0024] In Step 1 above n The mean of the data in the sampling period 、 、 The calculation formula is: ; ; ; in n Set according to the sampling period.

[0025] The calculation formula for the correlation coefficient between the three sensors in Step 2 above is: ; ; .

[0026] The deviation value of each correlation coefficient from the completely positive correlation within the sampling period in Step 2 above 、 and The calculation formula is: ; ; .

[0027] The specific method for the three-choose-two logic judgment in Step 3 above is: like ,and and , then the 3# sensor abnormality count is increased by 1, otherwise the 3# sensor abnormality count is set to 0; if the 3# sensor abnormality count value is equal to 5 n , then the output is 3# sensor fault; Similarly, if ,and and , then the abnormal count of sensor 2# is increased by 1, otherwise the abnormal count of sensor 2# is set to 0; if the abnormal count value of sensor 2# is equal to 5 n , then the output is 3# sensor fault; like ,and and , then the abnormal count of sensor 1# is increased by 1, otherwise the abnormal count of sensor 1# is set to 0; if the abnormal count value of sensor 1# is equal to 5 n , then the output 3# sensor failure.

[0028] In the above Step 3, if 、 、 Are less than or equal to , then 1#, 2#, 3# sensors are normal and have no faults; if 、 、 Both greater than , and the difference between the guide vane setting and the guide vane feedback is greater than the adjustment dead zone, the abnormal counts of 1#, 2#, and 3# sensors are all set to 1. When the abnormal count values ​​of the sensors are all equal to 5 n , then the output 1#, 2#, 3# sensor failure.

[0029] In the above Step 3 three-choose-two logic judgment, when 5 consecutive n Sampling period 、 、 When both are established and there is no deviation fault, the current main operation is maintained; 5 in a row n Sampling period 、 、 If all the above conditions are met, it is determined that the 1# sensor has a deviation fault, and the 2# sensor of the B set controller is switched to the main operation; 5 in a row n Sampling period 、 、 If all the above conditions are met, it is determined that the 2# sensor has a deviation fault, and the 1# sensor of the A set of controllers is switched to the main operation; 5 in a row nSampling period 、 、 If all of the above are true, it is determined that the 3# sensor has a deviation fault and the main function is maintained; 5 in a row n sampling period and the difference between the guide vane setting and the guide vane feedback is greater than the regulation dead zone, 、 、 If both are true, it is determined that the sensor is abnormal. At this time, it is necessary to use historical correlation data to determine which sensor is working properly and use it as the main sensor. T The total samples in time are N Historical data 、 、 Less than or equal to The number of samples N AB 、N AC 、N BC , calculate the correlation frequency between each pair of sensors 、 、 ; Then calculate the reliability frequency of a single sensor 、 、 ; Define sensor reliability weight W , for sensor A, , calculate the weighted deviation of the difference between sensor A and the other two sensors , calculate the comprehensive failure probability of sensor A Similarly, calculate the comprehensive failure probability of sensors B and C P B 、P C ,Comparing the sizes of the three, the smallest one is used as the main sensor, and the other two are marked as abnormal.

[0030] In the above Step 3, the two-choose-three logic judgment and the master selection judgment are carried out according to the following table: Three-choice-two primary selection decision table .

[0031] The movable parts of the above three sensors 1#, 2# and 3# are all connected to the transverse bracket and move with the relay. The transverse bracket is installed in the vertical direction of the connection between the relay piston rod and the control ring. The three sensors are arranged in parallel next to the relay.

[0032] Example 2: The Pearson correlation coefficient is a statistic that measures the degree of linear correlation between two variables. When the Pearson correlation coefficient is equal to 0, it means that the two variables are unrelated; when the Pearson correlation coefficient is greater than 0, it means that the two variables are positively correlated, and when it is equal to 1, it is completely positively correlated; when the Pearson correlation coefficient is less than 0, it means that the two variables are negatively correlated, and when it is equal to -1, it is completely negatively correlated. Three sensors are arranged in parallel next to the relay, and a transverse bracket is installed in the vertical direction at the connection between the relay piston rod and the control ring. The movable parts of the three sensors are all connected to the transverse bracket. Since they move with the relay, when the sensor has no fault and the output changes, the output value is close to completely positively correlated. Sensor fault judgment can be made by calculating the Pearson correlation coefficient between the output values ​​of each two sensors and the unit status. When all three sensors are working normally and there is a guide vane deviation, the correlation coefficient between the sensors is close to completely positive correlation, such as Figure 2-3 When one of the sensors has an over-limit, jump or dead value fault, and the other two sensors are working normally, the correlation coefficient between the faulty sensor and the other two sensors will deviate from a completely positive correlation, as shown in Figure 4-7 When two sensors fail or all three sensors fail, the correlation coefficients between the sensors will deviate from a completely positive correlation.

[0033] The specific fault diagnosis method is as follows: Step 1: Collect the code values ​​output by 3 sensors in real time , , . And calculate separately n The mean of the collected data 、 、 ,Right now 、 、 , n Set according to the sampling period, the value range is 5-10.

[0034] Step 2: Calculate the correlation coefficients between the three sensors 、 、 .by For example, the calculation formula is as follows: ; 、 The calculation of is the same as above. Calculate the deviation between each correlation coefficient of the current sampling period and the completely positive correlation, that is, , 、 Similarly. ,and and , then there is a strong correlation between 1# and 2# sensors, and the abnormal count of 3# sensor increases by 1. It is the abnormality judgment threshold, which is adjusted according to the accuracy of the sensor and is generally set to 0.001-0.05.

[0035] Step 3: Repeat step 2 to calculate the correlation coefficients for the next sampling period 、 、 , and determine the deviation value of each correlation coefficient from the completely positive correlation. If ,and and , then the 3# sensor abnormality count is increased by 1, otherwise the 3# sensor abnormality count is set to 0. If the 3# sensor abnormality count value is equal to 5 n , then the output is 3# sensor fault, and the same applies to 1# and 2# sensor fault judgment.

[0036] like 、 、 Are less than or equal to , then the 1#, 2#, and 3# sensors are normal and have no faults. 、 、 Both greater than , and the difference between the guide vane setting and the guide vane feedback is greater than the adjustment dead zone, the abnormal counts of 1#, 2#, and 3# sensors are all set to 1. When the abnormal count values ​​of the sensors are all equal to 5 n , then the output 1#, 2#, 3# sensor fault, sensor fault judgment process is as follows Figure 8-Figure 9 shown.

[0037] Simulate the dead value data of sensor 3# starting from the second sampling value. n Take 5 as an example to calculate the correlation coefficient between the three sensors, as shown in the following table and Figure 2 As shown, sensors 1# and 2# are almost completely positively correlated, while sensors 3# is negatively correlated with sensors 1# and 2#, which is inconsistent with the actual situation and indicates an anomaly.

[0038] 3 sensor sampling values ​​and correlation coefficients

[0039] Three-choice logic judgment: 5 consecutive n Sampling period 、 、 If all are established, there is no deviation fault and the current main operation is maintained; n Sampling period 、 、 If all of them are true, it is determined that the 1# sensor has a deviation fault, and the 2# sensor of the B machine is switched to the main operation; n Sampling period 、 、 If all are true, it is determined that the 2# sensor has a deviation fault, and the 1# sensor of the A machine is switched to the main operation; n Sampling period 、 、 If all are true, it is determined that the 3# sensor has a deviation fault and the main function is maintained; n sampling period and the difference between the guide vane setting and the guide vane feedback is greater than the regulation dead zone, 、 、 If both are true, it is determined that the sensor is abnormal. At this time, it is necessary to use historical correlation data to determine which sensor is working properly and use it as the main sensor. T The total samples in time are N Historical data 、 、 Less than or equal to The number of samples N AB 、N AC 、N BC , calculate the correlation frequency between each pair of sensors 、 、 ; Then calculate the reliability frequency of a single sensor 、 、 ; Define sensor reliability weight W , for sensor A, , calculate the weighted deviation of the difference between sensor A and the other two sensors , calculate the comprehensive failure probability of sensor A Similarly, calculate the comprehensive failure probability of sensors B and C P B 、P C , compare the sizes of the three, the smallest one is used as the main sensor, and the other two are marked as abnormal. The three-choose-two main selection judgment is shown in the table below.

[0040] Three-choice-two primary selection decision table .

Claims

1. A correlation-based control method for three-choose-two main displacement sensors, characterized in that the method include: Step 1: Real-time acquisition of the code values ​​output by three sensors 1#, 2#, and 3# , , , i represents the data collected in real time according to the collection cycle, and takes values ​​1, 2, 3...n; and calculates n The mean of the collected data 、 、 ;Step 2, calculate the distance between the three sensors n Correlation coefficient of sampling period 、 、 ; Calculate the deviation value of each correlation coefficient and the perfect positive correlation in the current sampling period respectively 、 and ;like ,and and , then there is a strong correlation between 1# and 2# sensors, and the abnormal count of 3# sensor increases by 1. This is the abnormality judgment threshold, which is adjusted according to the accuracy of the sensor; Step 3: Repeat Step 2 to calculate the correlation coefficients in the next sampling period. 、 、 , and determine the deviation value of each correlation coefficient from the complete positive correlation 、 and , and uses three-choose-two logic to determine the faulty sensor and select the sensor data to be adopted.

2. The correlation-based two-out-of-three control method for main contact displacement sensors according to claim 1, characterized in that: In Step 1 n The mean of the data in the sampling period 、 、 The calculation formula is: ； ； ； in n Set according to the sampling period.

3. The correlation-based two-out-of-three control method for main contact displacement sensors according to claim 2, characterized in that: The calculation formula for the correlation coefficient between the three sensors in Step 2 is: ； ； 。 4. The correlation-based two-out-of-three control method for main contact displacement sensors according to claim 3, characterized in that: The deviation value of each correlation coefficient from the completely positive correlation within the sampling period in Step 2 、 and The calculation formula is: ； ； 。 5. The correlation-based two-out-of-three control method for main contact displacement sensors according to claim 4, characterized in that: The specific method of the three-choose-two logic judgment in Step 3 is as follows: like ,and and , then the 3# sensor abnormality count is increased by 1, otherwise the 3# sensor abnormality count is set to 0; if the 3# sensor abnormality count value is equal to 5 n , then the output is 3# sensor fault; Similarly, if ,and and , then the abnormal count of sensor 2# is increased by 1, otherwise the abnormal count of sensor 2# is set to 0; if the abnormal count value of sensor 2# is equal to 5 n , then the output is 3# sensor fault; like ,and and , then the abnormal count of sensor 1# is increased by 1, otherwise the abnormal count of sensor 1# is set to 0; if the abnormal count value of sensor 1# is equal to 5 n , then the output 3# sensor failure.

6. The correlation-based two-out-of-three control method for main contact displacement sensors according to claim 5, characterized in that: In the above Step 3's three-choose-two logic judgment, if 、 、 Are less than or equal to , then 1#, 2#, 3# sensors are normal and have no faults; if 、 、 Both greater than , and the difference between the guide vane setting and the guide vane feedback is greater than the adjustment dead zone, the abnormal counts of 1#, 2#, and 3# sensors are all set to 1. When the abnormal count values ​​of the sensors are all equal to 5 n , then the output 1#, 2#, 3# sensor failure.

7. The correlation-based two-out-of-three control method for main contact displacement sensors according to claim 6, characterized in that: In the above Step 3, when there are 5 consecutive n Sampling period 、 、 When both are established and there is no deviation fault, the current main operation is maintained; 5 in a row n Sampling period 、 、 If all the above conditions are met, it is determined that the 1# sensor has a deviation fault, and the 2# sensor of the B set controller is switched to the main operation; 5 in a row n Sampling period 、 、 If all the above conditions are met, it is determined that the 2# sensor has a deviation fault, and the 1# sensor of the A set of controllers is switched to the main operation; 5 in a row n Sampling period 、 、 If all of the above are true, it is determined that the 3# sensor has a deviation fault and the main function is maintained; 5 in a row n sampling period and the difference between the guide vane setting and the guide vane feedback is greater than the regulation dead zone, 、 、 If both are true, it is determined that the sensor is abnormal. At this time, it is necessary to use historical correlation data to determine which sensor is working properly and use it as the main sensor. T The total samples in time are N Historical data 、 、 Less than or equal to The number of samples N AB 、N AC 、N BC , calculate the correlation frequency between each pair of sensors 、 、 ; Then calculate the reliability frequency of a single sensor 、 、 ; Define sensor reliability weight W , for sensor A, , calculate the weighted deviation of the difference between sensor A and the other two sensors , calculate the comprehensive failure probability of sensor A Similarly, calculate the comprehensive failure probability of sensors B and C P B 、P C ,Comparing the sizes of the three, the smallest one is used as the main sensor, and the other two are marked as abnormal.

8. The correlation-based two-out-of-three control method for main contact displacement sensors according to claim 7, characterized in that: In Step 3, the two-choose-three logic judgment and the master selection judgment are performed according to the following table: Three-choice-two primary selection decision table 。 9. The correlation-based two-out-of-three control method for main contact displacement sensors according to claim 8, characterized in that: The movable parts of the three sensors 1#, 2# and 3# are all connected to the transverse bracket and move with the relay. The transverse bracket is installed in the vertical direction of the connection between the relay piston rod and the control ring. The three sensors are arranged in parallel next to the relay.

Images