CN112113595B - Sensor fault detection method, device and computer readable medium - Google Patents

Sensor fault detection method, device and computer readable medium Download PDF

Info

Publication number
CN112113595B
CN112113595B CN202011020843.XA CN202011020843A CN112113595B CN 112113595 B CN112113595 B CN 112113595B CN 202011020843 A CN202011020843 A CN 202011020843A CN 112113595 B CN112113595 B CN 112113595B
Authority
CN
China
Prior art keywords
sensor
sensors
correlation coefficient
characterizing
correlation
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Active
Application number
CN202011020843.XA
Other languages
Chinese (zh)
Other versions
CN112113595A (en
Inventor
戴林杉
郑杰
韩三丰
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Siemens Mobility Technologies Beijing Co Ltd
Original Assignee
Siemens Mobility Technologies Beijing Co Ltd
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Siemens Mobility Technologies Beijing Co Ltd filed Critical Siemens Mobility Technologies Beijing Co Ltd
Priority to CN202011020843.XA priority Critical patent/CN112113595B/en
Publication of CN112113595A publication Critical patent/CN112113595A/en
Application granted granted Critical
Publication of CN112113595B publication Critical patent/CN112113595B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Classifications

    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01D—MEASURING NOT SPECIALLY ADAPTED FOR A SPECIFIC VARIABLE; ARRANGEMENTS FOR MEASURING TWO OR MORE VARIABLES NOT COVERED IN A SINGLE OTHER SUBCLASS; TARIFF METERING APPARATUS; MEASURING OR TESTING NOT OTHERWISE PROVIDED FOR
    • G01D18/00—Testing or calibrating apparatus or arrangements provided for in groups G01D1/00 - G01D15/00
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/22—Matching criteria, e.g. proximity measures

Landscapes

  • Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Data Mining & Analysis (AREA)
  • Theoretical Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Artificial Intelligence (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Evolutionary Biology (AREA)
  • Evolutionary Computation (AREA)
  • Bioinformatics & Computational Biology (AREA)
  • General Engineering & Computer Science (AREA)
  • Bioinformatics & Cheminformatics (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Testing And Monitoring For Control Systems (AREA)

Abstract

The invention provides a sensor fault detection method, a sensor fault detection device and a computer readable medium, wherein the sensor fault detection method comprises the following steps: acquiring historical readings of each sensor in the target automation control system; respectively determining a maximum support group of each sensor according to the acquired historical readings, wherein the maximum support group comprises at least one sensor which has correlation with the corresponding sensor; acquiring at least two current readings of each sensor in a current time window, wherein the current time window is a time period of which the end point time is the current time and the time span is equal to a preset time length; for each sensor, calculating the overall similarity of the sensor according to the current reading of the sensor and each sensor in the maximum support group of the sensor in the current time window; if the overall similarity of a sensor is less than a preset similarity threshold, the sensor is determined to be faulty. The scheme can be suitable for various automatic control systems to detect the sensor faults.

Description

Sensor fault detection method, device and computer readable medium
Technical Field
The present invention relates to the field of fault detection and analysis technologies, and in particular, to a method and an apparatus for detecting a sensor fault, and a computer readable medium.
Background
The sensors are indispensable parts in an automatic control system, various sensors acquire corresponding information and send the information to the controller, and the controller sends a control instruction to controlled equipment according to the information acquired by the sensors to realize automatic control, so that the accuracy of the information acquired by the sensors directly influences the reliability of the automatic control. In order to avoid the influence of the failure of the sensor on the reliability of the automation control system for the controlled device, the failure detection of the sensor in the automation control system is needed, and the failed sensor is discovered and repaired in time.
At present, a data model is generally constructed to detect faults of sensors, that is, a data model for an automatic control system is constructed, information acquired by each sensor in the automatic control system is input into the constructed data model, and whether a sensor has a fault is judged according to a result output by the data model.
Some sophisticated automation control systems include a large number of sensors and a large number of redundant sensors, such as Heating Ventilation and Air Conditioning (HVAC) systems on trains, which include a large number of temperature sensors, humidity sensors, and harmful gas concentration sensors, and a plurality of redundant sensors are provided for the same type of sensor. For the method for detecting the faults of the sensors by adopting the constructed data model at present, when an automatic control system comprises more types and number of the sensors and more redundant sensors, whether the sensors have faults or not can not be detected due to the difficulty in constructing the accurate data model, so that the applicability of the existing sensor fault detection method is poor.
Disclosure of Invention
In view of the above, the sensor fault detection method, the sensor fault detection device and the computer readable medium provided by the invention can be applied to various automatic control systems for sensor fault detection.
In a first aspect, an embodiment of the present invention provides a sensor fault detection method, including:
acquiring historical readings of each sensor in a target automation control system, wherein the target automation control system comprises at least two sensors;
for each sensor, determining a maximum support group of the sensor according to the acquired historical readings, wherein the maximum support group comprises at least one sensor which has correlation with the sensor;
acquiring at least two current readings of each sensor in a current time window, wherein the end time of the current time window is the current time, and the time span of the current time window is equal to a preset time length;
for each of the sensors, calculating an overall similarity for that sensor based on the current readings for that sensor and each of the sensors in the maximum supported set of that sensor within the current time window;
for each sensor, if the overall similarity of the sensor is smaller than a preset similarity threshold, determining that the sensor fails.
In a first possible implementation manner, with reference to the first aspect, for each sensor in the target automation control system, each sensor included in the maximum support set of the sensors is the same type of sensor as the sensor.
In a second possible implementation manner, with reference to the first aspect, the determining, for each sensor, a maximum support group of the sensor according to the obtained historical readings includes:
calculating a first correlation coefficient for characterizing the correlation between any two sensors of at least two sensors included in the automatic control system according to the similarity between the historical readings of the two sensors;
for each of the sensors, performing:
fitting the first correlation coefficient between the sensor and other sensors according to chi-square distribution to obtain a correlation distribution function;
determining a correlation coefficient threshold corresponding to a preset confidence probability threshold from the correlation distribution function;
determining the sensor with the first correlation coefficient larger than the correlation coefficient threshold value as a supporting sensor of the sensor;
determining a set of each of the supporting sensors as the maximum supporting set of the sensor.
In a third possible implementation manner, with reference to the first possible implementation manner, for any two sensors of at least two sensors included in the automatic control system, calculating a first correlation coefficient for characterizing a correlation between the two sensors according to a similarity between the historical readings of the two sensors, the method includes:
calculating the first correlation coefficient between the ith sensor and the jth sensor by a first formula set as follows for the ith sensor and the jth sensor of the at least two sensors included in the target automation control system;
the first formula set includes:
Figure GDA0002706275610000021
wherein d isijFor characterizing the first correlation coefficient between the ith and jth sensors, the uitFor characterizing the historical reading, u, of the ith sensor at the tth historical timejtThe current sensor is used for representing the historical reading of the jth sensor at the tth historical moment, the T is used for representing the total number of preset historical moments, and the T is an integer larger than 1.
In a fourth possible implementation manner, with reference to the first aspect and any one of the first possible implementation manner and the second possible implementation manner of the first aspect, for each sensor, calculating an overall similarity of the sensor according to the current readings of the sensor and each sensor in the maximum support group of the sensor in the current time window includes:
for each of the sensors, performing:
for each supporting sensor located in the maximum supporting group of the sensor, calculating a second correlation coefficient for characterizing a correlation between the supporting sensor and the sensor according to a similarity between the supporting sensor and the current reading of the sensor;
calculating the overall similarity of the sensor according to the second correlation coefficient between each supporting sensor in the maximum supporting group of the sensor and the sensor by the following formula;
Figure GDA0002706275610000031
wherein, R isiThe overall similarity for characterizing the i-th said sensor, said dipThe second correlation coefficient between the p-th and i-th said supported sensors in the maximum supported set for the i-th said sensor, the m being used to characterize the number of said supported sensors in the maximum supported set for the i-th said sensor.
In a fifth possible implementation manner, with reference to the third possible implementation manner, for each supporting sensor in the maximum supporting group of the sensor, calculating a second correlation coefficient for characterizing a correlation between the supporting sensor and the sensor according to a similarity between the current readings of the supporting sensor and the sensor, includes:
for each supporting sensor in the maximum support set of the sensor, calculating the second correlation coefficient between the supporting sensor and the sensor by a second set of equations;
the second set of equations includes:
Figure GDA0002706275610000032
wherein u isikFor characterizing the current reading, u, of the ith sensor at the kth acquisition instant within the current time windowpkThe current reading at the kth acquisition instant for the pth support sensor in the maximum supported group for the ith said sensor, the N is used to characterize the number of acquisition instants included within the current time window, and the N is an integer greater than 1.
In a second aspect, an embodiment of the present invention further provides a sensor fault detection apparatus, including:
the system comprises a historical reading acquisition module, a historical reading acquisition module and a historical reading acquisition module, wherein the historical reading acquisition module is used for acquiring the historical reading of each sensor in a target automation control system, and the target automation control system comprises at least two sensors;
a support group determination module, configured to determine, for each sensor, a maximum support group of the sensor according to the historical readings acquired by the historical reading acquisition module, where the maximum support group includes at least one sensor having a correlation with the sensor;
the current reading acquisition module is used for acquiring at least two current readings of each sensor in a current time window, wherein the end time of the current time window is the current time, and the time span of the current time window is equal to the preset time length;
a similarity calculation module, configured to calculate, for each sensor, an overall similarity of the sensor according to the current reading of the sensor in the current time window acquired by the current reading acquisition module and the current reading of each sensor in the maximum support group of the sensor determined by the support group determination module;
and the fault identification module is used for determining that the sensor has a fault when the similarity calculation module determines that the overall similarity of the sensor is smaller than a preset similarity threshold value for each sensor.
In a first possible implementation manner, in combination with the second aspect, for each sensor in the target automation control system, each sensor included in the maximum support set of sensors is the same type of sensor as the sensor.
In a second possible implementation manner, with reference to the second aspect, the support group determining module includes:
a first coefficient calculation sub-module for calculating, for any two of at least two sensors included in the automated control system, a first correlation coefficient characterizing a correlation between the two sensors based on a similarity between the historical readings of the two sensors;
a function fitting submodule for fitting the first correlation coefficient between the sensor and the other sensors calculated by the first coefficient calculation submodule according to chi-square distribution for each sensor to obtain a correlation distribution function;
a threshold value determining submodule for determining a correlation coefficient threshold value corresponding to a preset confidence probability threshold value from the correlation distribution function fitted by the function fitting submodule;
a sensor screening submodule, configured to determine, according to the correlation coefficient threshold determined by the threshold determination submodule, the sensor having the first correlation coefficient greater than the correlation coefficient threshold with respect to the corresponding sensor as a supporting sensor of the sensor;
a support group generation submodule for determining a set of the support sensors determined by the sensor screening submodule as the maximum support group of the sensor.
In a third possible implementation manner, in combination with the first possible implementation manner,
the first coefficient calculation sub-module is configured to calculate, for an ith sensor and a jth sensor of at least two sensors included in the target automation control system, the first correlation coefficient between the ith sensor and the jth sensor by using a first formula set as follows;
the first formula set includes:
Figure GDA0002706275610000051
wherein d isijFor characterizing the first correlation coefficient between the ith and jth sensors, the uitFor characterizing the historical reading, u, of the ith sensor at the tth historical timejtFor characterizing the historical reading of the jth sensor at the tth historical timeThe T is used for representing the total number of preset historical moments and is an integer greater than 1.
In a fourth possible implementation manner, with reference to the second aspect and any one of the first possible implementation manner and the second possible implementation manner of the second aspect, the similarity calculation module includes:
a second coefficient calculation sub-module for calculating, for any one of said sensors and each supporting sensor located in said maximum supporting set of sensors, a second correlation coefficient characterizing a correlation between the supporting sensor and the sensor based on a similarity between the current readings of the supporting sensor and the sensor;
a similarity operator module for calculating, for each sensor, the overall similarity of the sensor according to the second correlation coefficient between each of the supporting sensors in the maximum supporting group of the sensor and the sensor calculated by the second coefficient calculation sub-module;
Figure GDA0002706275610000052
wherein, R isiThe overall similarity for characterizing the i-th said sensor, said dipThe second correlation coefficient between the p-th and i-th said supported sensors in the maximum supported set for the i-th said sensor, the m being used to characterize the number of said supported sensors in the maximum supported set for the i-th said sensor.
In a fifth possible implementation manner, in combination with the third possible implementation manner,
the second coefficient calculation submodule is used for calculating the second correlation coefficient between any one sensor and each supporting sensor in the maximum supporting group of the sensor through the following second formula group;
the second set of equations includes:
Figure GDA0002706275610000061
wherein u isikFor characterizing the current reading, u, of the ith sensor at the kth acquisition instant within the current time windowpkThe current reading at the kth acquisition instant for the pth support sensor in the maximum supported group for the ith said sensor, the N is used to characterize the number of acquisition instants included within the current time window, and the N is an integer greater than 1.
In a third aspect, an embodiment of the present invention further provides another sensor fault detection apparatus, including: the method comprises the following steps: at least one memory and at least one processor;
the at least one memory to store a machine readable program;
the at least one processor is configured to invoke the machine-readable program to perform the method provided by the first aspect and any possible implementation manner of the first aspect.
In a fourth aspect, the present invention further provides a computer-readable medium, where computer instructions are stored, and when executed by a processor, cause the processor to perform the method provided by the first aspect and any possible implementation manner of the first aspect.
According to the technical scheme, the maximum support group corresponding to each sensor can be respectively determined according to the historical reading of each sensor in the target automatic control system, such that each sensor's largest supported group includes sensors that have a correlation with that sensor, when a sensor fails the reading of that sensor will differ significantly from the readings of the sensors in its maximum support group, after at least two current readings of each sensor in the current time window are obtained, the overall similarity of the sensor is calculated according to the current readings of each sensor and each sensor in the maximum support group of the sensor, and the calculated overall similarity characterizes the similarity of readings between the corresponding sensor and each sensor in its maximum support set, and then whether the corresponding sensor fails or not can be determined according to the calculated overall similarity. Therefore, the maximum support group of each sensor is respectively determined according to the historical reading of each sensor in the automatic control system, then the overall similarity of each sensor is respectively calculated according to the reading of each sensor and each sensor in the maximum support group in the latest period of time, and whether the corresponding sensor has a fault or not can be further determined according to the overall similarity, so that the fault detection of the sensors in the automatic control system can be realized without constructing a data model of the automatic control system, the fault detection of the sensors can be realized for simple and complex automatic control systems, and the applicability of the sensor fault detection method can be improved.
Drawings
FIG. 1 is a flow chart of a method for sensor fault detection provided by one embodiment of the present invention;
fig. 2 is a flowchart of a maximum supported group determining method according to an embodiment of the present invention;
FIG. 3 is a flow chart of a method for overall similarity calculation according to an embodiment of the present invention;
FIG. 4 is a schematic diagram of a sensor failure detection arrangement provided in accordance with an embodiment of the present invention;
FIG. 5 is a schematic diagram of another sensor failure detection arrangement provided in accordance with an embodiment of the present invention;
FIG. 6 is a schematic diagram of yet another sensor failure detection arrangement provided in accordance with an embodiment of the present invention;
fig. 7 is a schematic diagram of another sensor failure detection apparatus according to an embodiment of the present invention.
List of reference numerals:
11: obtaining historical readings for each sensor in a target automation control system
12: for each sensor, determining the maximum support group of the sensor according to the acquired historical readings
13: obtaining at least two current readings of each sensor within a current time window
14: respectively calculating the overall similarity of each sensor according to each maximum support group and the current reading of each sensor
15: determining that a sensor is malfunctioning when the overall similarity of the sensor is less than a preset similarity threshold
121: calculating a first correlation coefficient between any two sensors based on historical readings of the two sensors
122: respectively fitting a correlation distribution function to first correlation coefficients between each sensor and other sensors
123: determining a threshold value of the correlation coefficient from the correlation distribution function of each sensor
124: determining a sensor corresponding to the first correlation coefficient being greater than the correlation coefficient threshold as a supporting sensor
125: determining a set of supporting sensors as a maximum supporting group of respective sensors
141: calculating a second number of correlations between the sensor and its supporting sensors based on the respective current readings
142: calculating the overall similarity of a sensor according to the second correlation number of the sensor and each supporting sensor
10: sensor failure detection method 20: sensor failure detection device 30: sensor fault detection device
21: the history data reading module 22: the support group determination module 23: current reading acquisition module
24: the similarity calculation module 25: the fault identification module 221: first coefficient calculation submodule
222: function fitting submodule 223: sensor screening submodule 224: support group generation submodule
241: the second coefficient calculation sub-module 242: similarity operator module 31: memory device
32: processor with a memory having a plurality of memory cells
Detailed Description
As described above, when detecting a fault of a sensor in an automation control system, a data model for the automation control system needs to be constructed in advance, and then information collected by the sensor is input into the constructed data model during the operation of the automation control system, and whether the sensor in the automation control system has a fault is determined according to a result output by the data model. For the automatic control systems with fewer types and numbers of sensors, a data model corresponding to the automatic control system can be conveniently constructed, and then the constructed data model is used for judging whether the sensors have faults or not, but for the automatic control systems with higher types and numbers of sensors, such as a heating ventilation and air conditioning system on a train, which comprises a large number of redundant sensors, the data model corresponding to the automatic control system is difficult to construct, so that the fault detection of the sensors in the automatic control system by using the data model cannot be realized.
In the embodiment of the invention, for an automatic control system comprising a large number of redundant sensors, correlation exists between different sensors in the automatic control system, each sensor can determine a supporting sensor which has correlation with the sensor according to historical readings of the sensors, when one sensor fails, the reading of the sensor and the reading of each supporting sensor are not synchronized, so that the overall similarity can be calculated according to the reading of the sensor and each supporting sensor, the similarity of the readings between the sensor and the supporting sensor is represented through the overall similarity, and whether the corresponding sensor fails or not can be judged according to the overall similarity. Therefore, the supporting sensor of each sensor can be determined according to the historical reading of each sensor in the automatic control system, the overall similarity of each sensor can be calculated according to the reading of each sensor and the reading of the supporting sensor of each sensor, and whether the sensor fails or not is determined according to the calculated overall similarity, so that the fault detection of the sensors can be realized without constructing a data model of the automatic control system, and the method can be suitable for various automatic control systems to detect the faults of the sensors.
The following describes a sensor failure detection method and apparatus provided by an embodiment of the present invention in detail with reference to the accompanying drawings.
As shown in fig. 1, an embodiment of the present invention provides a sensor fault detection method 10, which may include the steps of:
step 11: acquiring historical readings of each sensor in a target automation control system, wherein the target automation control system comprises at least two sensors;
step 12: for each sensor, determining a maximum support group of the sensor according to the acquired historical readings, wherein the maximum support group comprises at least one sensor which is related to the sensor;
step 13: acquiring at least two current readings of each sensor in a current time window, wherein the end time of the current time window is the current time, and the time span of the current time window is equal to a preset time length;
step 14: for each sensor, calculating the overall similarity of the sensor according to the current readings of the sensor and each sensor in the maximum support group of the sensor in the current time window;
step 15: for each sensor, if the overall similarity of the sensor is less than a preset similarity threshold, determining that the sensor is faulty.
In the embodiment of the present invention, the maximum support group corresponding to each sensor may be determined according to the historical readings of the sensors in the target automation control system, so that the maximum support group of each sensor includes the sensor having a correlation with the sensor, that is, each sensor in the maximum support group has a correlation with the sensor. When one sensor fails, the reading of the sensor and the reading of each sensor in the maximum support group of the sensor generate obvious asynchronous difference, after at least two current readings of each sensor in the current time window are obtained, the overall similarity of the sensor is calculated according to the current readings of each sensor and each sensor in the maximum support group of the sensor, the calculated overall similarity represents the reading similarity between the corresponding sensor and each sensor in the maximum support group of the sensor, and then whether the corresponding sensor fails or not can be determined according to the calculated overall similarity. Therefore, the maximum support group of each sensor is respectively determined according to the historical reading of each sensor in the automatic control system, then the overall similarity of each sensor is respectively calculated according to the reading of each sensor and each sensor in the maximum support group in the latest period of time, and whether the corresponding sensor has a fault or not can be further determined according to the overall similarity, so that the fault detection of the sensors in the automatic control system can be realized without constructing a data model of the automatic control system, the fault detection of the sensors can be realized for simple and complex automatic control systems, and the applicability of the sensor fault detection method can be improved.
In the embodiment of the present invention, the step 11 and the step 12 are used to respectively determine the maximum support group of each sensor in the target automation control system, and since the correlation between readings of different sensors in the automation control system usually does not change, after the maximum support group of each sensor is determined, multiple fault detections may be performed on each sensor according to the determined maximum support group, and it is not necessary to re-determine the maximum support group every time a fault detection is performed on a sensor, so that the fault detection of the sensor can be rapidly achieved.
When sensors are newly added, the number of sensors is reduced, the layout positions of the sensors are changed, or all or part of the sensors are recalibrated in the target automation control system, the steps 11 and 12 can be executed again to update the maximum support group of each sensor in the target automation control system, and it is ensured that each sensor has strong correlation with the reading of each sensor in the maximum support group.
Unlike steps 11 and 12, steps 13 to 15 are repeated periodically, for example, a sensor failure detection period may be preset, and steps 13 to 15 are sequentially executed once after every set sensor failure detection period to determine whether each sensor in the target automatic control system has failed.
In the embodiment of the present invention, for each sensor in the target automation control system, the maximum support group of the sensor includes at least one other sensor having a correlation with the sensor, where the correlation specifically refers to the correlation between the sensor and the readings of the sensors in the maximum support group. Since at least one sensor is included in the maximum support group, when two or more sensors are included in the maximum support group of one sensor, the overall similarity of the sensor can be determined according to the reading of the sensor and each sensor in the maximum support group of the sensor in the last period of time, so that whether the sensor is in fault or not can be determined according to the reading of the sensor and at least two other sensors, and the accuracy of fault identification of the sensor can be ensured.
In the embodiment of the present invention, when obtaining the historical readings of each sensor in the target automation control system in step 11, the readings of each sensor in a specified time period may be used as the historical readings, and the most recent N (N is a positive integer) readings of each sensor may also be used as the historical readings. Specifically, a historical time window may be specified, and then readings of the sensors in the automation control system within the specified historical time window may be obtained from the historical database as historical readings, and a sample number may also be specified, and a most recent sample number of readings of each sensor in the target automation control system may be used as historical readings. For example, at least one reading of each sensor in the HVAC system between a first time in the past and a second time in the past may be determined as a historical reading when detecting a fault with the sensor in the HVAC system, or the last 20000 readings of each sensor in the HVAC system may also be determined as historical readings. According to different types of sensors, the historical readings and the current readings of the sensors have different significances, for example, the historical readings and the current readings of the temperature sensor are temperature values, the historical readings and the current readings of the pressure sensor are pressure values, the historical readings and the current readings of the humidity sensor are humidity values, and the historical readings and the current readings of the harmful gas concentration sensor are harmful gas concentrations.
It should be noted that, when obtaining the historical readings of each sensor in the target automation control system, it is necessary to ensure that each sensor in the target automation control system is in a normal operating state, so as to ensure that the maximum support group of each sensor can be accurately determined according to the obtained historical readings.
In the embodiment of the present invention, in order to calculate the overall similarity of each sensor, at least two current readings of each sensor in the target automation control system in the current time window may be obtained, and then the overall similarity of each sensor may be calculated according to the current readings of each sensor and each sensor in the maximum support group thereof, respectively. The current time window refers to a time period in which the time span is equal to the preset time length and the end time is the current time, that is, the current time window is the latest time, so that the current time window is dynamically changed along with the lapse of time. The time span of the current time window can be determined according to the frequency of data collected by each sensor, the current time window with smaller time span can be adopted when the frequency of data collected by each sensor is higher, and the current time window with larger time span can be adopted when the frequency of data collected by each sensor is lower.
In the embodiment of the present invention, for each sensor, when calculating the overall similarity of the sensor, at least two current readings of each sensor in the sensor and the maximum support group of the sensor are respectively obtained, and then the overall similarity of the sensor is calculated according to the at least two current readings of the sensor and the at least two current readings of each sensor in the maximum support group of the sensor. Therefore, the overall similarity of each sensor is calculated through the readings of the sensors at least two moments, so that the sensor is prevented from influencing the fault detection result of the sensor due to the fact that the single accidental factor causes the fluctuation of the readings, and the accuracy of detecting the fault of the sensor is guaranteed.
In the embodiment of the invention, corresponding similarity threshold values are preset for each sensor, after the overall similarity of one sensor is calculated, the calculated overall similarity is compared with the similarity threshold value of the sensor, if the overall similarity of the sensor is smaller than the similarity threshold value of the sensor, the large asynchronous deviation of the readings of the sensor and the sensors in the maximum support group occurs, and the readings of the sensor and the sensors in the maximum support group should be synchronously changed under normal conditions, so that the sensor can be determined to have a fault.
It should be noted that, because the parameters detected by different sensors are different and the arrangement positions of different sensors are also different, the similarity threshold needs to be preset for each sensor, so that the similarity thresholds corresponding to different sensors may be different. The similarity threshold can be manually specified by experts in related fields, and can also be obtained by analyzing historical readings of the sensor through a machine learning algorithm.
Optionally, on the basis of the sensor failure detection method 10 shown in fig. 1, step 12 determines the maximum support group corresponding to each sensor, so that each sensor has correlation with each sensor in its maximum support group, and further, any sensor and each sensor in its maximum support group may be the same type of sensor.
In the embodiment of the invention, for the automatic control system comprising a large number of redundant sensors, the same parameter information is detected by a plurality of sensors of the same type, and automatic control is carried out according to the parameter information acquired by each sensor, so that the automatic control process is prevented from being influenced by the fault of one sensor, and the reliability of the automatic control system is ensured. For each sensor, a set of a plurality of sensors which have the same type and correlation with the sensor is used as the maximum support group of the sensor, and the sensor and each sensor in the maximum support group thereof are the same type of sensor, so that the readings of the sensor and each sensor in the maximum support group thereof reflect the same parameters, the correlation between the sensor and each sensor in the maximum support group thereof is closer and more obvious, the subsequently calculated overall similarity can be ensured to accurately reflect whether the reading of the sensor and each sensor in the maximum support group thereof are asynchronously changed, and the accuracy of fault detection of the sensor can be improved.
For example, the target automation control system includes a temperature sensor and a humidity sensor, the maximum support group corresponding to each temperature sensor includes one or more temperature sensors having a correlation with the temperature sensor, and the maximum support group corresponding to each humidity sensor includes one or more humidity sensors having a correlation with the humidity sensor. Further, in the HVAC system, the temperature sensor a, the temperature sensor B and the temperature sensor C are arranged in the vicinity of the top of the same train car, and readings of the three temperature sensors are changed synchronously when the temperature in the train car is changed under normal conditions, so that the three temperature sensors have correlation with each other, the temperature sensor B and the temperature sensor C are included in the maximum support group of the temperature sensor a, the temperature sensor a and the temperature sensor C are included in the maximum support group of the temperature sensor B, and the temperature sensor a and the temperature sensor B are included in the maximum support group of the temperature sensor C.
Alternatively, on the basis of the sensor fault detection method 10 shown in fig. 1, when the step 12 determines the maximum support group of each sensor, for each sensor, first correlation coefficients for characterizing the similarity between the sensor and each other sensor may be calculated, and then the maximum support group of the sensor is determined according to the calculated first correlation coefficients. As shown in fig. 2, the method of determining the maximum supported group may include the steps of:
step 121: calculating a first correlation coefficient for representing the correlation between any two sensors in at least two sensors included in the automation control system according to the similarity between historical readings of the two sensors;
step 122: for each sensor, fitting a first correlation coefficient between the sensor and other sensors according to chi-square distribution to obtain a correlation distribution function;
step 123: for each sensor, determining a correlation coefficient threshold corresponding to a preset confidence probability threshold from a correlation distribution function of the sensor;
step 124: for each sensor, determining a sensor with a first correlation coefficient between the sensor and a correlation coefficient threshold value as a supporting sensor of the sensor;
step 125: for each sensor, a set of supporting sensors for that sensor is determined as the maximum supported set of that sensor.
In the embodiment of the invention, each sensor in the target automation control system is firstly combined pairwise, and aiming at any two combined sensors, calculating a first correlation coefficient characterizing a correlation between the two sensors based on a similarity between the two sensor historical readings, wherein the similarity between the historical readings of the two sensors is represented by the degree of synchronization of the historical readings of the two sensors over time, i.e. the first correlation coefficient characterizes a synchronization procedure of historical readings of the two sensors over time, if the historical readings of two sensors change over time in a higher synchronization procedure, the first correlation coefficient between the two sensors is larger, whereas if the historical readings of the two sensors change over time with a lower synchronization procedure or change out of synchronization, the first correlation coefficient between the two sensors is smaller.
For example, in a historical time period, the temperature values collected by the temperature sensor A and the temperature sensor B are synchronously increased and decreased, there is a high similarity between the historical readings of temperature sensor a and temperature sensor B, the first correlation coefficient between the respective temperature sensors a and B is large, while the temperature values collected by the temperature sensor A and the temperature sensor C have no synchronous change rule, namely, when the temperature value collected by the temperature sensor A is increased, the temperature value collected by the temperature sensor C is increased or decreased, when the temperature value collected by the temperature sensor A is decreased, the temperature value collected by the temperature sensor C is also increased or decreased, there is a lower similarity between the historical readings of temperature sensor a and temperature sensor C and the corresponding first correlation coefficient between temperature sensor a and temperature sensor C is smaller.
When the first correlation coefficient is calculated according to the historical readings, the average value of the difference values of the readings of the two sensors at each historical moment can be used as the first correlation coefficient between the two sensors, or the average value of the historical readings of each sensor can be calculated first, and then the first correlation coefficient is calculated according to the deviation of the historical readings of each sensor and the average value of the historical readings. As for the second method for calculating the first correlation coefficient, the following embodiments will be described in detail.
In the embodiment of the invention, after the first correlation coefficient between one sensor and each other sensor is determined, the first correlation coefficients between the sensor and each other sensor are fitted according to a chi-square distribution to obtain a correlation distribution function corresponding to the sensor, then a correlation coefficient threshold corresponding to a preset confidence probability threshold is determined from the correlation distribution function of the sensor, then the sensor with the first correlation coefficient between the sensors larger than the determined correlation coefficient threshold is determined as a supporting sensor of the sensor, and finally a set consisting of each supporting sensor of the sensor is determined as a maximum supporting set of the sensor.
For each sensor, after calculating a first correlation coefficient between the sensor and each other sensor, fitting the first correlation coefficient between the sensor and each other sensor according to a chi-square distribution to obtain a correlation distribution function corresponding to the sensor, where the correlation distribution function represents a distribution of the first correlation coefficient between the sensor and each other sensor, and the first correlation coefficient represents a correlation between the sensor and the corresponding sensor, so that after determining a correlation coefficient threshold corresponding to a preset confidence probability threshold from the correlation distribution function, a sensor with a first correlation coefficient between the sensor and the corresponding sensor greater than the correlation coefficient threshold can be used as an element in a maximum support group of the sensor.
For each sensor, fitting first correlation coefficients between the sensor and other sensors according to chi-square distribution to obtain a correlation distribution function, accurately determining a supporting sensor with a strong correlation between a reading and the sensor according to the correlation distribution function, and determining a set consisting of all supporting sensors of the sensor as a maximum supporting group of the sensor.
In the embodiment of the present invention, after the first correlation coefficients between one sensor and other sensors are calculated, the calculated first correlation coefficients are fitted according to a chi-square distribution, so as to obtain the following correlation distribution function:
Figure GDA0002706275610000131
wherein d isijFor characterizing a first correlation coefficient between the ith sensor and the jth sensor, n for characterizing a total number of sensors comprised by the target automation control system,
Figure GDA0002706275610000132
for characterizing the gamma function.
After the correlation distribution function is fitted, a correlation coefficient threshold corresponding to a preset confidence probability threshold is determined from the correlation distribution function, and the threshold probability threshold may be set to be 95% in a normal case.
Alternatively, on the basis of the maximum support group determination method shown in fig. 2, when the step 121 calculates the first correlation coefficient between the two sensors, the correlation metric or the distance metric may be used as the first correlation coefficient. The correlation metric may be Pearson correlation coefficient (Pearson correlation coefficient), spearman's rank correlation coefficient (pearman's rank correlation coefficient), Kendall rank correlation coefficient (Kendall rank correlation coefficient), etc., the Distance metric may be Euclidean Distance (Euclidean Distance) or Manhattan Distance (Manhattan Distance), etc., and specifically, the correlation metric or the Distance metric may be used as the first correlation coefficient according to actual requirements.
When the pearson correlation coefficient is selected as the first correlation coefficient, for the ith sensor and the jth sensor in the target automation control system, the correlation coefficient between the ith sensor and the jth sensor can be calculated by a first formula set as follows;
wherein the first formula set includes:
Figure GDA0002706275610000141
wherein d isijFor characterizing a first correlation coefficient, u, between an ith sensor and a jth sensor in a target automation control systemitFor characterizing the historical reading, u, of the ith sensor at the tth historical timejtThe method is used for representing the historical reading of the jth sensor at the tth historical moment, T is used for representing the total number of preset historical moments, and T is an integer greater than 1.
In the embodiment of the invention, a historical time window is preset, T historical moments are included in the historical time window, and when historical readings of all sensors in a target automation control system are obtained, the readings of all the sensors at all the historical moments in the historical time window are respectively obtained as the historical readings.
In the embodiment of the invention, after the historical readings of each sensor are acquired, the historical readings of any two sensors are substituted into the first formula group, the first correlation coefficient between the two sensors is calculated, and the calculated first correlation coefficient can accurately represent the correlation between the readings of the two sensors. And respectively substituting the historical readings of different sensors into the first formula group to calculate a first correlation coefficient between any two sensors.
Alternatively, on the basis of the sensor failure detection method 10 shown in fig. 1, when the overall similarity of the sensors is calculated in step 14, the second correlation coefficient between one sensor and each supporting sensor in the maximum supporting group of the sensor may be calculated first, and then the overall similarity of the sensor may be calculated according to the calculated second correlation coefficients. As shown in fig. 3, the method of calculating the overall similarity of a sensor may include the steps of:
step 141: for each supporting sensor in the maximum supporting group of the sensors, calculating a second correlation coefficient for characterizing the correlation between the supporting sensor and the sensor according to the similarity between the supporting sensor and the current reading of the sensor;
step 142: calculating the overall similarity of the sensor according to a second correlation coefficient between each supporting sensor in the maximum supporting group of the sensors and the sensor by the following formula;
Figure GDA0002706275610000151
wherein R isiFor characterizing the overall similarity of the ith sensor, dipA second correlation coefficient between the p-th supported sensor and the i-th sensor in the maximum supported set for the i-th sensor, and m is used to characterize the number of supported sensors in the maximum supported set for the i-th sensor.
In the embodiment of the present invention, similar to the way of calculating the first correlation coefficient by using the historical readings, after determining the maximum support group of each sensor and acquiring the current reading of each sensor, for each support sensor in the maximum support group of one sensor, a second correlation coefficient between the sensor and the support sensor may be calculated according to the current readings of the sensor and the support sensor, so that the second correlation represents the correlation between the sensor and the support sensor.
In the embodiment of the invention, after the second correlation coefficients between one sensor and each supporting sensor in the maximum supporting group of the sensor are calculated, the average value of the second correlation coefficients calculated for the sensor is calculated by the formula based on Bayesian estimation to serve as the overall similarity of the sensor, so that the calculated overall similarity can reflect the difference between the current reading of the sensor and the current reading of each supporting sensor as a whole, and whether the sensor fails or not can be judged according to the calculated overall similarity.
It should be noted that, the overall similarity of the sensors is calculated through the above formula, and since the calculation mode and the calculation process are both simple, the overall similarity of each sensor can be quickly calculated, thereby ensuring the timeliness of fault detection of each sensor. Of course, in addition to calculating the overall similarity of the sensors by the above formula, the overall similarity of the sensors may be calculated by other manners, for example, the overall similarity of the sensors may be calculated by the following formula:
Figure GDA0002706275610000152
wherein R isiFor characterizing the overall similarity of the ith sensor, dipA second correlation coefficient between the p-th supported sensor and the i-th sensor in the maximum support set for characterizing the i-th sensor, m is a number of supported sensors in the maximum support set for characterizing the i-th sensor, (mu)0，σ0 2) For characterizing the support data of the ith sensor on itself in a Bayesian estimation, (mu)p，σp 2) For characterizing the support data of the pth support sensor on itself in Bayesian estimation, μ0For characterizing the mean, sigma, of data collected by the ith sensor0 2For characterizing the variance, mu, of the data collected by the ith sensorpFor characterizing the mean, sigma, of the data collected by the pth sensorp 2For characterizing the variance of the data collected by the pth support sensor.
When the overall similarity of each sensor is calculated through the formula, although the calculation process is complex and the timeliness of the detection of the sensors is reduced, the calculated overall similarity is more accurate due to the introduction of the support data of the sensors and the support sensors for the sensors.
Alternatively, on the basis of the overall similarity calculation method shown in fig. 3, when the step 141 calculates the second correlation coefficient between one sensor and one of its supporting sensors, the correlation metric or the distance metric may be used as the second correlation coefficient. The correlation metric may be Pearson correlation coefficient (Pearson correlation coefficient), Spearman correlation coefficient (Spearman's rank correlation coefficient), Kendall rank correlation coefficient (Kendall rank correlation coefficient), etc., the Distance metric may be Euclidean Distance (Euclidean Distance) or Manhattan Distance (Manhattan Distance), etc., and the correlation metric or the Distance metric may be used as the first correlation coefficient according to actual requirements.
In a more preferred embodiment, in order to ensure that the calculated overall similarity can reflect the read correlation between the sensor and each of its supporting sensors more accurately, the first correlation coefficient and the second correlation coefficient may be calculated in the same manner when determining the maximum supporting group and calculating the overall similarity, for example, the first correlation coefficient and the second correlation coefficient both use pearson correlation coefficient, or the first correlation coefficient and the second correlation coefficient both use euclidean distance.
When the pearson correlation coefficient is selected as the second correlation coefficient, for each supporting sensor located in the maximum supporting group of one sensor, the second correlation coefficient between the supporting sensor and the sensor may be calculated by the following second formula group;
wherein the second set of equations comprises:
Figure GDA0002706275610000161
wherein u isikFor characterizing the current reading, u, of the ith sensor at the kth acquisition instant in the current time windowpkThe current reading of the p-th support sensor in the maximum support group for the ith sensor at the k-th acquisition time is characterized, and N is used for characterizing the current timeThe number of acquisition instants included in the inter-window, and N is an integer greater than 1.
In the embodiment of the invention, the current time window comprises N acquisition moments, and when the current reading of each sensor in the target automation control system is acquired, the reading of each sensor at each acquisition moment in the current time window is acquired as the current reading.
In the embodiment of the present invention, after obtaining the current reading of each sensor, the current reading of one sensor and one supporting sensor thereof is substituted into the second formula set, so as to calculate the second correlation coefficient between the sensor and the supporting sensor, and the calculated second correlation coefficient can accurately represent the correlation between the sensor and the supporting sensor in reading. By substituting the current readings of each sensor and its supporting sensor into the second formula set, a second correlation coefficient between any one sensor and any one supporting sensor can be calculated.
It should be noted that the maximum support group of each sensor is respectively determined according to the historical reading of each sensor in the automatic control system, the overall similarity of each sensor is respectively calculated according to the current reading of each sensor and each sensor in the maximum support group of each sensor, the overall similarity represents the asynchronous difference between the corresponding sensor and each sensor in the maximum support group of the corresponding sensor in reading, and whether the corresponding sensor fails or not can be judged according to the calculated overall similarity. The embodiment of the invention adopts the idea of sensor fusion and realizes the detection of the sensor fault according to the difference of the readings of the sensors with correlation. Because the maximum support group is determined according to the historical reading of each sensor, no expert definition is needed, the sensor fault detection method 10 is easier to realize in real time, and after the sensors in the automatic control system are changed, the corresponding maximum support group can be determined for each sensor again, and the method has strong applicability.
It should be noted that, when the sensor fault detection method 10 provided by the embodiments of the present invention is applied to an HVAC system to perform sensor fault detection, since the HVAC system includes a temperature sensor, a humidity sensor and a harmful gas concentration sensor, the acquired historical readings and current readings may each include a temperature value, a humidity value and a harmful gas concentration value.
As shown in fig. 4, one embodiment of the present invention provides a sensor failure detection apparatus 20 including:
a historical reading acquisition module 21, configured to acquire a historical reading of each sensor in the target automation control system, where the target automation control system includes at least two sensors;
a support group determining module 22, configured to determine, for each sensor, a maximum support group of the sensor according to the historical readings acquired by the historical reading acquiring module 21, where the maximum support group includes at least one sensor having a correlation with the sensor;
a current reading obtaining module 23, configured to obtain at least two current readings of each sensor in a current time window, where an end time of the current time window is a current time, and a time span of the current time window is equal to a preset time length;
a similarity calculation module 24, configured to calculate, for each sensor, an overall similarity of the sensor according to the current readings of the sensor in the current time window acquired by the current reading acquisition module 23 and the sensors in the maximum support group of the sensor determined by the support group determination module 22;
and a fault identification module 25, configured to determine, for each sensor, that the sensor is faulty when the similarity calculation module 24 determines that the overall similarity of the sensor is smaller than a preset similarity threshold.
In an embodiment of the present invention, the historical reading obtaining module 21 may be configured to perform step 11 in the above-described method embodiment, the support group determining module 22 may be configured to perform step 12 in the above-described method embodiment, the current reading obtaining module 23 may be configured to perform step 13 in the above-described method embodiment, the similarity calculating module 24 may be configured to perform step 14 in the above-described method embodiment, and the fault identifying module 25 may be configured to perform step 15 in the above-described method embodiment.
Alternatively, on the basis of the sensor failure detection device 20 shown in fig. 4, for each sensor in the target automation control system, each sensor included in the maximum support group of the sensors is the same type of sensor as the sensor.
Alternatively, on the basis of the sensor failure detection apparatus 20 shown in fig. 4, as shown in fig. 5, the support group determination module 22 includes:
a first coefficient calculation sub-module 221, configured to calculate, for any two sensors of at least two sensors included in the automation control system, a first correlation coefficient characterizing a correlation between the two sensors according to a similarity between historical readings of the two sensors;
a function fitting submodule 222, configured to fit, for each sensor, the first correlation coefficient between the sensor and the other sensors calculated by the first coefficient calculation submodule 221 according to a chi-square distribution, so as to obtain a correlation distribution function;
a threshold determination submodule 223 for determining a correlation coefficient threshold corresponding to the preset confidence probability threshold from the correlation distribution function fitted by the function fitting submodule 222;
a sensor screening submodule 224, configured to determine, according to the correlation coefficient threshold determined by the threshold determination submodule 223, a sensor, for which a first correlation coefficient with a corresponding sensor is greater than the correlation coefficient threshold, as a supporting sensor of the sensor;
a support group generation submodule 225 for determining the set of support sensors determined by the sensor screening submodule 224 as the maximum support group of the sensor.
In an embodiment of the present invention, the first coefficient calculation sub-module 221 may be configured to perform step 121 in the above-described method embodiment, the function fitting sub-module 222 may be configured to perform step 122 in the above-described method embodiment, the threshold determination sub-module 223 may be configured to perform step 123 in the above-described method embodiment, the sensor screening sub-module 224 may be configured to perform step 124 in the above-described method embodiment, and the support group generation sub-module 225 may be configured to perform step 125 in the above-described method embodiment.
Alternatively, on the basis of the sensor failure detection apparatus 20 shown in fig. 5, the first coefficient calculation sub-module 221 is configured to calculate, for an ith sensor and a jth sensor of the at least two sensors included in the target automation control system, a first correlation coefficient between the ith sensor and the jth sensor by using a first formula set as follows;
the first formula set includes:
Figure GDA0002706275610000181
wherein d isijFor characterizing a first correlation coefficient, u, between the ith and jth sensorsitFor characterizing the historical reading, u, of the ith sensor at the tth historical timejtThe method is used for representing the historical reading of the jth sensor at the tth historical moment, T is used for representing the total number of preset historical moments, and T is an integer greater than 1.
Alternatively, on the basis of the sensor failure detection device 20 shown in fig. 4, as shown in fig. 6, the similarity calculation module 24 includes:
a second coefficient calculation submodule 241 for calculating, for any sensor and each support sensor located in the maximum support group of sensors, a second correlation coefficient characterizing the correlation between the support sensor and the sensor based on the similarity between the current readings of the support sensor and the sensor;
a similarity operator module 242, configured to calculate, for each sensor, the overall similarity of the sensor according to the second correlation coefficient between each supporting sensor in the maximum supporting group of the sensor and the sensor calculated by the second coefficient calculating sub-module 241, by using the following formula;
Figure GDA0002706275610000191
wherein R isiFor characterizing the overall similarity of the ith sensor, dipA second correlation coefficient between the p-th supported sensor and the i-th sensor in the maximum supported set for the i-th sensor, and m is used to characterize the number of supported sensors in the maximum supported set for the i-th sensor.
In the embodiment of the present invention, the second coefficient calculating submodule 241 may be configured to perform step 141 in the above-described method embodiment, and the similarity degree operator module 242 may be configured to perform step 142 in the above-described method embodiment.
Alternatively, on the basis of the sensor failure detection device shown in fig. 6, the second coefficient calculation submodule 241 is configured to calculate, for any one sensor and each support sensor located in the maximum support group of the sensor, a second correlation coefficient between the support sensor and the sensor by the following second formula group;
the second formula set includes:
Figure GDA0002706275610000192
wherein u isikFor characterizing the current reading, u, of the ith sensor at the kth acquisition instant in the current time windowpkA current reading at a kth acquisition instant of a pth support sensor in the maximum support group for the ith sensor, N is used to characterize a number of acquisition instants included within the current time window, and N is an integer greater than 1.
As shown in fig. 7, another sensor failure detection apparatus 30 according to an embodiment of the present invention includes: at least one memory 31 and at least one processor 32;
the at least one memory 31 for storing a machine readable program;
the at least one processor 32 is configured to invoke the machine readable program to execute the sensor failure detection method provided in the above embodiments.
It should be noted that, because the contents of information interaction, execution process, and the like between the modules and sub-modules in the sensor fault detection device 20/30 are based on the same concept as the foregoing method embodiment, specific contents may refer to the description in the foregoing method embodiment, and are not described herein again.
The present invention also provides a computer readable medium storing instructions for causing a computer to perform a sensor fault detection method as herein described. Specifically, a system or an apparatus equipped with a storage medium on which software program codes that realize the functions of any of the above-described embodiments are stored may be provided, and a computer (or a CPU or MPU) of the system or the apparatus is caused to read out and execute the program codes stored in the storage medium.
In this case, the program code itself read from the storage medium can realize the functions of any of the above-described embodiments, and thus the program code and the storage medium storing the program code constitute a part of the present invention.
Examples of the storage medium for supplying the program code include a floppy disk, a hard disk, a magneto-optical disk, an optical disk (e.g., CD-ROM, CD-R, CD-RW, DVD-ROM, DVD-RAM, DVD-RW, DVD + RW), a magnetic tape, a nonvolatile memory card, and a ROM. Alternatively, the program code may be downloaded from a server computer via a communications network.
Further, it should be clear that the functions of any one of the above-described embodiments may be implemented not only by executing the program code read out by the computer, but also by causing an operating system or the like operating on the computer to perform a part or all of the actual operations based on instructions of the program code.
Further, it is to be understood that the program code read out from the storage medium is written to a memory provided in an expansion board inserted into the computer or to a memory provided in an expansion unit connected to the computer, and then causes a CPU or the like mounted on the expansion board or the expansion unit to perform part or all of the actual operations based on instructions of the program code, thereby realizing the functions of any of the above-described embodiments.
It should be noted that not all steps and modules in the above flows and system structure diagrams are necessary, and some steps or modules may be omitted according to actual needs. The execution order of the steps is not fixed and can be adjusted as required. The system structure described in the above embodiments may be a physical structure or a logical structure, that is, some modules may be implemented by the same physical entity, or some modules may be implemented by a plurality of physical entities, or some components in a plurality of independent devices may be implemented together.
In the above embodiments, the hardware unit may be implemented mechanically or electrically. For example, a hardware element may comprise permanently dedicated circuitry or logic (such as a dedicated processor, FPGA or ASIC) to perform the corresponding operations. The hardware elements may also comprise programmable logic or circuitry, such as a general purpose processor or other programmable processor, that may be temporarily configured by software to perform the corresponding operations. The specific implementation (mechanical, or dedicated permanent, or temporarily set) may be determined based on cost and time considerations.
While the invention has been shown and described in detail in the drawings and in the preferred embodiments, it is not intended to limit the invention to the embodiments disclosed, and it will be apparent to those skilled in the art that various combinations of the code auditing means in the various embodiments described above may be used to obtain further embodiments of the invention, which are also within the scope of the invention.

Claims (14)

1. A sensor fault detection method (10), comprising:
acquiring historical readings of each sensor in a target automation control system, wherein the target automation control system comprises at least two sensors;
for each sensor, determining a maximum support group of the sensor according to the obtained historical readings, wherein the maximum support group of the sensor comprises at least one sensor which has correlation with the sensor, and the maximum support group of the sensor does not comprise the sensor; the sensor changes synchronously with the readings of each sensor in its maximum support group;
acquiring at least two current readings of each sensor in a current time window, wherein the end time of the current time window is the current time, and the time span of the current time window is equal to a preset time length;
for each of the sensors, calculating an overall similarity for the sensor based on the current readings for the sensor and the sensors in the maximum supported group of sensors within the current time window;
for each sensor, if the overall similarity of the sensor is smaller than a preset similarity threshold, determining that the sensor fails.
2. The method of claim 1,
for each of the sensors, the sensors included in the maximum support group of the sensor are of the same type as the sensor.
3. The method of claim 1, wherein determining, for each of the sensors, a maximum supported set of the sensor based on the obtained historical readings comprises:
calculating a first correlation coefficient for characterizing the correlation between any two sensors of at least two sensors included in the automatic control system according to the similarity between the historical readings of the two sensors;
for each of the sensors, performing:
fitting the first correlation coefficient between the sensor and other sensors according to chi-square distribution to obtain a correlation distribution function;
determining a correlation coefficient threshold corresponding to a preset confidence probability threshold from the correlation distribution function;
determining a sensor with the first correlation coefficient between the sensor and the sensor larger than the correlation coefficient threshold value as a supporting sensor of the sensor;
determining a set of each of the supporting sensors as the maximum supporting set of the sensor.
4. The method of claim 3, wherein calculating, for any two of at least two sensors included in the automated control system, a first correlation coefficient characterizing a correlation between the two sensors based on a similarity between the historical readings of the two sensors comprises:
for a second of the at least two sensors comprised by the target automation control system
Figure DEST_PATH_IMAGE001
A sensor and
Figure 799143DEST_PATH_IMAGE002
a sensor for calculating the second equation by the following first equation group
Figure 52532DEST_PATH_IMAGE001
A sensor and the second
Figure 560874DEST_PATH_IMAGE002
The first correlation coefficient between the sensors;
the first formula set includes:
Figure 786581DEST_PATH_IMAGE004
wherein, the
Figure DEST_PATH_IMAGE005
For characterizing said
Figure 128701DEST_PATH_IMAGE001
A sensor and the second
Figure 72649DEST_PATH_IMAGE002
The first correlation coefficient between the sensors, the
Figure 650260DEST_PATH_IMAGE006
For characterizing in
Figure DEST_PATH_IMAGE007
At the time of history
Figure 917425DEST_PATH_IMAGE001
The historical readings of the individual sensors are taken,
Figure 520793DEST_PATH_IMAGE008
for characterizing in
Figure 155299DEST_PATH_IMAGE007
At the time of history
Figure 864498DEST_PATH_IMAGE002
The historical readings of individual sensors, the
Figure DEST_PATH_IMAGE009
For characterizing a predetermined total number of historical moments, and
Figure 189431DEST_PATH_IMAGE009
is an integer greater than 1.
5. The method of any one of claims 1 to 4, wherein said calculating, for each of said sensors, an overall similarity for that sensor based on said current readings for that sensor and each sensor in said maximum supported set of that sensor within said current time window comprises:
for each of the sensors, performing:
for each supporting sensor located in the maximum supporting group of the sensor, calculating a second correlation coefficient for characterizing a correlation between the supporting sensor and the sensor according to a similarity between the supporting sensor and the current reading of the sensor;
calculating the overall similarity of the sensor according to the second correlation coefficient between each supporting sensor in the maximum supporting group of the sensor and the sensor by the following formula;
Figure DEST_PATH_IMAGE011
wherein, the
Figure 578081DEST_PATH_IMAGE012
For characterizing
Figure 995155DEST_PATH_IMAGE001
The overall similarity of each of the sensors, the
Figure DEST_PATH_IMAGE013
For characterizing
Figure 806248DEST_PATH_IMAGE001
The first of the maximum support group of the sensors
Figure 126633DEST_PATH_IMAGE014
The support sensor and the second
Figure 637249DEST_PATH_IMAGE001
The second correlation coefficient between the sensors, the
Figure DEST_PATH_IMAGE015
For characterizing
Figure 109994DEST_PATH_IMAGE001
A number of the supporting sensors in the maximum supported set of the sensors.
6. The method of claim 5, wherein calculating, for each supporting sensor in the maximum supporting set of sensors, a second correlation coefficient characterizing a correlation between the supporting sensor and the sensor based on a similarity between the supporting sensor and the current reading of the sensor comprises:
for each supporting sensor in the maximum support set of the sensor, calculating the second correlation coefficient between the supporting sensor and the sensor by a second set of equations;
the second set of equations includes:
Figure DEST_PATH_IMAGE017
wherein, the
Figure 255936DEST_PATH_IMAGE018
For characterizing
Figure 726100DEST_PATH_IMAGE001
The first sensor in the current time window
Figure DEST_PATH_IMAGE019
The current reading at each of the acquisition times,
Figure 971399DEST_PATH_IMAGE020
for characterizing
Figure 864531DEST_PATH_IMAGE001
The first of the maximum support group of the sensors
Figure 922486DEST_PATH_IMAGE014
The support sensor is arranged at the second position
Figure 748622DEST_PATH_IMAGE019
The current reading at each acquisition time, the
Figure DEST_PATH_IMAGE021
For characterizing the number of acquisition instants comprised within said current time window, and
Figure 945248DEST_PATH_IMAGE021
is an integer greater than 1.
7. Sensor failure detection apparatus (20) comprising:
a historical reading acquisition module (21) for acquiring historical readings of each sensor in the target automation control system, wherein the target automation control system comprises at least two sensors;
a support group determination module (22) for determining, for each of the sensors, a maximum support group of the sensor according to the historical readings acquired by the historical reading acquisition module (21), wherein the maximum support group of the sensor includes at least one sensor having a correlation with the sensor, and the maximum support group of the sensor does not include the sensor; the sensor changes synchronously with the readings of each sensor in its maximum support group;
a current reading obtaining module (23) for obtaining at least two current readings of each sensor in a current time window, wherein the end time of the current time window is the current time, and the time span of the current time window is equal to a preset time length;
a similarity calculation module (24) for calculating, for each of the sensors, an overall similarity of the sensor based on the current readings of the sensor in the current time window acquired by the current reading acquisition module (23) and the sensors in the maximum support group of the sensor determined by the support group determination module (22);
a fault identification module (25) for determining, for each of the sensors, that the sensor is faulty when the similarity calculation module (24) determines that the overall similarity of the sensor is less than a preset similarity threshold.
8. The apparatus of claim 7,
for each of the sensors, the sensors included in the maximum support group of the sensor are of the same type as the sensor.
9. The apparatus of claim 7, wherein the support group determination module (22) comprises:
a first coefficient calculation submodule (221) for calculating, for any two of at least two sensors comprised by said automatic control system, a first correlation coefficient characterizing a correlation between the two sensors, on the basis of a similarity between said historical readings of the two sensors;
a function fitting submodule (222) for fitting the first correlation coefficient between the sensor and the other sensors calculated by the first coefficient calculation submodule (221) according to a chi-square distribution for each of the sensors to obtain a correlation distribution function;
a threshold determination submodule (223) for determining a correlation coefficient threshold corresponding to a predetermined confidence probability threshold from the correlation distribution function fitted by the function fitting submodule (222);
a sensor screening submodule (224) for determining, as a sensor supporting the sensor, a sensor having the first correlation coefficient with the corresponding sensor larger than the correlation coefficient threshold value, based on the correlation coefficient threshold value determined by the threshold value determining submodule (223);
a support group generation submodule (225) for determining a set of said support sensors determined by said sensor screening submodule (224) as said maximum support group for said sensor.
10. The apparatus of claim 9,
the first coefficient calculation submodule (221) is used for aiming at the second of at least two sensors included in the target automation control system
Figure 466621DEST_PATH_IMAGE001
A sensor and
Figure 282262DEST_PATH_IMAGE002
a sensor for calculating the second equation by the following first equation group
Figure 664701DEST_PATH_IMAGE001
A sensor and the second
Figure 279284DEST_PATH_IMAGE002
The first correlation coefficient between the sensors;
the first formula set includes:
Figure 114385DEST_PATH_IMAGE022
wherein, the
Figure 546765DEST_PATH_IMAGE005
For watchesIs characterized by
Figure 783712DEST_PATH_IMAGE001
A sensor and the second
Figure 72873DEST_PATH_IMAGE002
The first correlation coefficient between the sensors, the
Figure 192007DEST_PATH_IMAGE006
For characterizing in
Figure 896920DEST_PATH_IMAGE007
At the time of history
Figure 926056DEST_PATH_IMAGE001
The historical readings of the individual sensors are taken,
Figure 792643DEST_PATH_IMAGE008
for characterizing in
Figure 399074DEST_PATH_IMAGE007
At the time of history
Figure 907678DEST_PATH_IMAGE002
The historical readings of individual sensors, the
Figure 384796DEST_PATH_IMAGE009
For characterizing a predetermined total number of historical moments, and
Figure 687863DEST_PATH_IMAGE009
is an integer greater than 1.
11. The apparatus according to any one of claims 7 to 10, wherein the similarity calculation module (24) comprises:
a second coefficient calculation submodule (241) for calculating, for any one of said sensors and each support sensor located in said maximum support set of sensors, a second correlation coefficient characterizing the correlation between the support sensor and the sensor based on the similarity between the support sensor and said current reading of the sensor;
a similarity operator module (242) for calculating, for each sensor, said overall similarity of the sensor according to said second correlation coefficient between each said supporting sensor of said maximum supporting group of the sensor and the sensor calculated by said second coefficient calculation submodule (241);
Figure 250432DEST_PATH_IMAGE011
wherein, the
Figure 297148DEST_PATH_IMAGE012
For characterizing
Figure 894351DEST_PATH_IMAGE001
The overall similarity of each of the sensors, the
Figure 887760DEST_PATH_IMAGE013
For characterizing
Figure 937624DEST_PATH_IMAGE001
The first of the maximum support group of the sensors
Figure 788031DEST_PATH_IMAGE014
The support sensor and the second
Figure 662577DEST_PATH_IMAGE001
A said transmitterThe second correlation coefficient between sensors, the
Figure 805982DEST_PATH_IMAGE015
For characterizing
Figure 844608DEST_PATH_IMAGE001
A number of the supporting sensors in the maximum supported set of the sensors.
12. The apparatus of claim 11,
the second coefficient calculating submodule (241) is used for calculating the second correlation coefficient between any one sensor and each supporting sensor in the maximum supporting group of the sensor through the following second formula group;
the second set of equations includes:
Figure DEST_PATH_IMAGE023
wherein, the
Figure 154497DEST_PATH_IMAGE018
For characterizing
Figure 460714DEST_PATH_IMAGE001
The first sensor in the current time window
Figure 276485DEST_PATH_IMAGE019
The current reading at each of the acquisition times,
Figure 474511DEST_PATH_IMAGE020
for characterizing
Figure 271914DEST_PATH_IMAGE001
The first of the maximum support group of the sensors
Figure 245686DEST_PATH_IMAGE014
The support sensor is arranged at the second position
Figure 465315DEST_PATH_IMAGE019
The current reading at each acquisition time, the
Figure 72008DEST_PATH_IMAGE021
For characterizing the number of acquisition instants comprised within said current time window, and
Figure 848334DEST_PATH_IMAGE021
is an integer greater than 1.
13. A sensor failure detection device (30), characterized by comprising: at least one memory (31) and at least one processor (32);
the at least one memory (31) for storing a machine readable program;
the at least one processor (32) configured to invoke the machine readable program to perform the method of any of claims 1 to 6.
14. A computer readable medium having stored thereon computer instructions which, when executed by a processor, cause the processor to perform the method of any of claims 1 to 6.
CN202011020843.XA 2020-09-25 2020-09-25 Sensor fault detection method, device and computer readable medium Active CN112113595B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202011020843.XA CN112113595B (en) 2020-09-25 2020-09-25 Sensor fault detection method, device and computer readable medium

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202011020843.XA CN112113595B (en) 2020-09-25 2020-09-25 Sensor fault detection method, device and computer readable medium

Publications (2)

Publication Number Publication Date
CN112113595A CN112113595A (en) 2020-12-22
CN112113595B true CN112113595B (en) 2021-03-30

Family

ID=73801662

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202011020843.XA Active CN112113595B (en) 2020-09-25 2020-09-25 Sensor fault detection method, device and computer readable medium

Country Status (1)

Country Link
CN (1) CN112113595B (en)

Families Citing this family (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN114414023A (en) * 2021-12-30 2022-04-29 北京万集科技股份有限公司 Sensor abnormality diagnosis method and device, and storage medium
CN116484306B (en) * 2023-06-20 2023-09-26 蘑菇物联技术（深圳）有限公司 Positioning method and device of abnormal sensor, computer equipment and storage medium
CN117419758A (en) * 2023-12-18 2024-01-19 潍柴动力股份有限公司 Engine sensor fault detection method, device, equipment and readable storage medium

Citations (8)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN104517027A (en) * 2013-10-07 2015-04-15 横河电机株式会社 State diagnosing method and state diagnosing apparatus
CN106096634A (en) * 2016-06-06 2016-11-09 北京理工大学 Fault detection method based on Adaptive windowing mental arithmetic method with interval halving algorithm
CN108180935A (en) * 2018-01-31 2018-06-19 深圳春沐源控股有限公司 Fault detection method and device of sensor
KR20180106701A (en) * 2017-03-21 2018-10-01 (주)심플랫폼 Device management system and method based on Internet Of Things
CN108981781A (en) * 2017-05-30 2018-12-11 迪尔公司 For analyzing and detecting the hypothesis analysis system and method for machine sensor fault
CN111126632A (en) * 2019-11-25 2020-05-08 合肥美的电冰箱有限公司 Household appliance fault prediction method and device, refrigerator and storage medium
CN111260893A (en) * 2020-01-10 2020-06-09 中国海洋石油集团有限公司 Fault early warning method and device for ocean platform propeller
CN111555162A (en) * 2020-04-06 2020-08-18 连云港市港圣开关制造有限公司 High-efficient intelligent power distribution operation and maintenance system

Patent Citations (8)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN104517027A (en) * 2013-10-07 2015-04-15 横河电机株式会社 State diagnosing method and state diagnosing apparatus
CN106096634A (en) * 2016-06-06 2016-11-09 北京理工大学 Fault detection method based on Adaptive windowing mental arithmetic method with interval halving algorithm
KR20180106701A (en) * 2017-03-21 2018-10-01 (주)심플랫폼 Device management system and method based on Internet Of Things
CN108981781A (en) * 2017-05-30 2018-12-11 迪尔公司 For analyzing and detecting the hypothesis analysis system and method for machine sensor fault
CN108180935A (en) * 2018-01-31 2018-06-19 深圳春沐源控股有限公司 Fault detection method and device of sensor
CN111126632A (en) * 2019-11-25 2020-05-08 合肥美的电冰箱有限公司 Household appliance fault prediction method and device, refrigerator and storage medium
CN111260893A (en) * 2020-01-10 2020-06-09 中国海洋石油集团有限公司 Fault early warning method and device for ocean platform propeller
CN111555162A (en) * 2020-04-06 2020-08-18 连云港市港圣开关制造有限公司 High-efficient intelligent power distribution operation and maintenance system

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
基于定性趋势分析的空调系统传感器故障检测;宋政辉 等;《计算机测量与控制》;20181231;第30页第0节-第33页第4.1节，图1、3 *

Also Published As

Publication number Publication date
CN112113595A (en) 2020-12-22

Similar Documents

Publication Publication Date Title
CN112113595B (en) Sensor fault detection method, device and computer readable medium
Hauer et al. Did we test all scenarios for automated and autonomous driving systems?
US6181975B1 (en) Industrial process surveillance system
US5463768A (en) Method and system for analyzing error logs for diagnostics
EP1950634B1 (en) Process for adapting measurement suite configuration for gas turbine performance diagnostics
EP2778818B1 (en) Identification of faults in a target system
CN110288007B (en) Data labeling method and device and electronic equipment
JP2004118839A (en) Method for supporting specification of function unit failed in technical equipment
CN102566421B (en) The system and method for the conditional dependencies modeling of abnormality detection in machine state monitoring
US20200143292A1 (en) Signature enhancement for deviation measurement-based classification of a detected anomaly in an industrial asset
EP2405357A2 (en) Methods systems and apparatus for determining whether built-in-test fault codes are indicative of an actual fault condition or a false alarm
KR20170031985A (en) Fault detection and diagnostics method of air-conditioning system
CN115168868B (en) Business vulnerability analysis method and server applied to artificial intelligence
CN112771462B (en) Degradation detection system
CN111639798A (en) Intelligent prediction model selection method and device
CN111124732A (en) Disk fault prediction method, system, device and storage medium
CN105446314A (en) Sensor system using safety mechanism
JP5911581B2 (en) Logic drawing error location estimation apparatus and method
US20210138873A1 (en) Deterioration diagnosis device, deterioration diagnosis method and computer readable medium
US20050102566A1 (en) Method for diagnosing complex system faults
CN111368782B (en) Training method of coal fire area recognition model, and coal fire area recognition method and device
US20220215138A1 (en) Method for Validating System Parameters of an Energy System, Method for Operating an Energy System, and Energy Management System for an Energy System
WO2021031091A1 (en) Method and apparatus for detecting operating status of device, and computer readable medium
WO2021046723A1 (en) Sensor model generation method and system, and sensor measurement method and system
CN113269268A (en) Data matching method and system based on big data and cloud platform

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination
GR01 Patent grant
GR01 Patent grant