CN110045716B - Method and system for detecting and diagnosing early fault of closed-loop control system - Google Patents
Method and system for detecting and diagnosing early fault of closed-loop control system Download PDFInfo
- Publication number
- CN110045716B CN110045716B CN201910312767.0A CN201910312767A CN110045716B CN 110045716 B CN110045716 B CN 110045716B CN 201910312767 A CN201910312767 A CN 201910312767A CN 110045716 B CN110045716 B CN 110045716B
- Authority
- CN
- China
- Prior art keywords
- control system
- loop control
- closed
- wavelet transform
- residual
- 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
Links
Images
Classifications
-
- G—PHYSICS
- G05—CONTROLLING; REGULATING
- G05B—CONTROL OR REGULATING SYSTEMS IN GENERAL; FUNCTIONAL ELEMENTS OF SUCH SYSTEMS; MONITORING OR TESTING ARRANGEMENTS FOR SUCH SYSTEMS OR ELEMENTS
- G05B23/00—Testing or monitoring of control systems or parts thereof
- G05B23/02—Electric testing or monitoring
- G05B23/0205—Electric testing or monitoring by means of a monitoring system capable of detecting and responding to faults
- G05B23/0259—Electric testing or monitoring by means of a monitoring system capable of detecting and responding to faults characterized by the response to fault detection
- G05B23/0262—Confirmation of fault detection, e.g. extra checks to confirm that a failure has indeed occurred
-
- G—PHYSICS
- G05—CONTROLLING; REGULATING
- G05B—CONTROL OR REGULATING SYSTEMS IN GENERAL; FUNCTIONAL ELEMENTS OF SUCH SYSTEMS; MONITORING OR TESTING ARRANGEMENTS FOR SUCH SYSTEMS OR ELEMENTS
- G05B2219/00—Program-control systems
- G05B2219/20—Pc systems
- G05B2219/24—Pc safety
- G05B2219/24065—Real time diagnostics
Abstract
The invention discloses a method for detecting and diagnosing early faults of a closed-loop control system, which comprises the following steps: s0, establishing a real on-site closed-loop control system according to actual requirements; s1, establishing a virtual closed-loop control system based on a data assimilation technology and a mathematical mechanism model of the equipment; s2, analyzing and obtaining the residual error of the real closed-loop control system relative to the virtual numerical control system based on the real closed-loop control system and the virtual closed-loop control system; and S3, identifying the fault type in the real closed-loop control system according to the obtained residual error. The invention solves the technical problems of low diagnosis accuracy and difficult diagnosis of the tiny faults in the prior art, and can realize the quick and accurate identification of the tiny faults in the closed-loop control system.
Description
Technical Field
The invention relates to a fault detection technology suitable for a closed-loop control system of a nuclear power plant, in particular to a method and a system for detecting and diagnosing early faults of the closed-loop control system.
Background
Modern control systems and devices are increasingly complex and scaled up, and such systems can cause significant loss of life and property in the event of failure. Fault diagnosis is becoming increasingly important as an important method and powerful measure to improve system reliability and reduce the risk of accidents. However, regardless of how large and aggressive the fault is, the faults start from minor faults (early, incipient). Early minor fault detection faces the following difficulties: (1) the external interference causes difficulty in diagnosing a minor fault. The operation environment of the equipment is complex and changeable, and the micro fault has weak symptoms, so that the whole system cannot be seriously influenced in the early stage, and all main operation parameters are within an acceptable design range. (2) The closed loop control structure compensates for the effects of faults on the equipment and results in the propagation and evolution of minor faults. The control system of the nuclear power plant generally adopts a closed-loop control structure, and has considerable fault tolerance capability for faults within a certain grade range. The failure occurs automatically, and the system can not generate great performance reduction, namely, controller 'compensation effect', and the compensation effect can cause the monitoring system of the train to generate a great amount of failure report for early tiny failures.
In the prior art, one of the technical solutions is a fault diagnosis method based on a Symbol Directed Graph (SDG), which is a diagnosis technique based on a qualitative model. The basic idea of the method is as follows: the method comprises the steps of describing system causal behaviors of a system in a normal or fault state by using an SDG model, capturing useful information to finish fault diagnosis according to an established causal relationship diagram, and finding out a fault source and a development and evolution process of a fault in the process along a branch direction by combining a certain search strategy. And further provides a fault diagnosis framework which combines SDG multiplication and Qualitative Trend Analysis (QTA) to diagnose early faults. Under this framework, SDG is level 1, providing a set of possible candidates for failure; and 2, according to the change of the current time value measured by the sensor, diagnosing early faults by using the QTA. The second technical scheme in the prior art is an online approximation method, and the basic idea of diagnosing the micro fault by the online approximation method (OLA) is as follows: and designing an adaptive estimator according to an adaptive theory. The self-adaptive rate of the parameters can be adjusted on line by designing, so that the constructed model is matched with the original system with enough precision. When the output of the on-line approximation fault function is not zero, it means that a fault has occurred. In actual engineering, faults are not all represented as external linear faults, therefore, for a class of nonlinear early faults with amplitude coefficients changing in a negative exponential rule, OLA is designed to capture nonlinear characteristics of the faults, a learning method for early fault detection is established, and in the framework, any deviation caused by the faults in the system is monitored by a nonlinear adaptive fault estimator, so that fault diagnosis is realized.
In practical application, the technology at least has the following technical problems: in the first technical scheme, the number of nodes, the number of branches of the SDG, and the complex relationship between the nodes and the branches increase with the complexity of the system, and the increase of the number of nodes and the number of branches increases the heavy task of modeling the SDG and the reasoning burden of diagnosing faults by the SDG, so that when the fault of the complex system is diagnosed by using the SDG method, the situations of poor real-time performance and low diagnosis accuracy rate occur; SDG node thresholds are difficult to determine accurately, which is crucial for diagnosing minor faults; the same fault symptom adopts more than one fault reason inferred by the SDG, and the inferred fault reasons have the same possibility and are difficult to further distinguish. In the second technical scheme, the design of the self-adaptive estimator is difficult to realize in engineering; and the fault needs to be estimated through an online self-adaptive estimator, and the estimation result can be used for fault separation and fault identification, but the accuracy of fault estimation cannot be theoretically proved at present.
Based on this, it is urgently needed to find a new fault detection and diagnosis method for accurately detecting and diagnosing a minor fault in a closed-loop control system.
Disclosure of Invention
The method and the system for detecting and diagnosing the early faults of the closed-loop control system solve the technical problem that the micro faults cannot be identified due to the structural compensation or fault tolerance of the closed-loop control in the prior art, and achieve the purposes of high detection efficiency, high speed and more accordance with actual field requirements.
In one aspect, the present application provides a method for early fault detection and diagnosis of a closed-loop control system, the method comprising the steps of:
s0, establishing a real on-site closed-loop control system according to requirements; s1, establishing a virtual closed-loop control system based on a data assimilation technology and a mathematical mechanism model of the equipment; s2, analyzing and obtaining a residual error of the real closed-loop control system relative to the virtual closed-loop control system based on the real closed-loop control system and the virtual closed-loop control system; and S3, identifying the fault type in the real closed-loop control system according to the obtained residual error.
Optionally, the step S2 includes: s21, analyzing and obtaining a first detection deviation value in the field real closed-loop control system according to the real closed-loop control system; s22, analyzing and obtaining a second detection deviation value in the virtual closed-loop control system according to the virtual closed-loop control system; and S23, acquiring a residual error of the real closed-loop control system relative to the virtual closed-loop control system according to the first detection deviation value and the second detection deviation value.
Optionally, the step S3 includes: s31, carrying out multi-scale and multi-resolution refinement analysis on the obtained residual error by utilizing wavelet transform, and obtaining a continuous wavelet transform coefficient of the residual error through discretization; s32, constructing a residual continuous wavelet transform coefficient fuzzy rule base according to the residual continuous wavelet transform coefficients; and S33, acquiring the fault type in the real closed-loop control system according to the residual continuous wavelet transform coefficient fuzzy rule base.
Step S32 further includes: s321, calculating the average value of the residual continuous wavelet transform coefficients, and taking the average value as a threshold; s322, calculating the sum of the residual continuous wavelet transform coefficients, calculating the difference value of the threshold value and the sum of the residual continuous wavelet transform coefficients, and dividing the sum of the residual continuous wavelet transform coefficients into a limited range according to the difference value and through a membership function; s323, according to the limited range of the sum of the residual continuous wavelet transform coefficients, correspondingly establishing limited fault types, and constructing a residual continuous wavelet transform coefficient fuzzy rule base.
Optionally, the wavelet used in the wavelet transform is a mexica-hat wavelet.
In another aspect, the present application further provides a closed-loop control system early failure detection and diagnosis system, including:
the real closed-loop control system establishing module is used for establishing a field real closed-loop control system according to requirements; the virtual numerical control system establishing module is connected with the real closed-loop control system establishing module and used for establishing a virtual closed-loop control system based on a data assimilation technology and a mathematical mechanism model of equipment; the residual error analysis and acquisition module is connected with the real closed-loop control system and the virtual numerical control system and is used for analyzing and acquiring the residual error of the real closed-loop control system relative to the virtual numerical control system based on the real closed-loop control system and the virtual numerical control system; and the fault identification module is connected with the residual error analysis and acquisition module and used for identifying the fault type in the real closed-loop control system according to the obtained residual error.
Optionally, the residual analysis and acquisition module includes: the first detection deviation value acquisition module is connected with the real closed-loop control system establishment module and used for analyzing and acquiring a first detection deviation value in the field real closed-loop control system according to the real closed-loop control system; the second detection deviation value acquisition module is connected with the virtual numerical control system establishment module and used for analyzing and acquiring a second detection deviation value in the virtual numerical closed-loop control system; and the residual error acquisition module is connected with the first detection deviation value acquisition module and the second detection deviation value acquisition module and is used for acquiring the residual error of the real closed-loop control system relative to the virtual closed-loop control system according to the first detection deviation value and the second detection deviation value.
Optionally, the fault identifying module includes: the continuous wavelet transform coefficient acquisition module is connected with the residual error acquisition module and used for carrying out multi-scale and multi-resolution refinement analysis on the obtained residual error by utilizing wavelet transform and obtaining a residual error continuous wavelet transform coefficient through discretization; the fuzzy rule base establishing module is connected with the continuous wavelet transform coefficient acquiring module and used for establishing a residual continuous wavelet transform coefficient fuzzy rule base according to the residual continuous wavelet transform coefficients; and the fault type identification module is connected with the continuous wavelet transform coefficient acquisition module and the fuzzy rule base establishment module and is used for acquiring the fault type in the real closed-loop control system according to the residual continuous wavelet transform coefficient fuzzy rule base.
Optionally, the fuzzy rule base establishing module includes: the threshold module is connected with the continuous wavelet transform coefficient acquisition module and used for calculating the average value of the residual continuous wavelet transform coefficients and taking the average value as a threshold; the segmentation module is connected with the continuous wavelet transform coefficient acquisition module and the threshold module and used for calculating the sum of the residual continuous wavelet transform coefficients, calculating the difference value of the threshold and the sum of the residual continuous wavelet transform coefficients, and dividing the sum of the residual continuous wavelet transform coefficients into a limited range according to the difference value and through a membership function; and the fuzzy rule base rule establishing module is connected with the continuous wavelet transform coefficient acquisition module and the segmentation module and is used for correspondingly establishing limited fault types according to the limited range of the sum of the residual continuous wavelet transform coefficients and establishing a residual continuous wavelet transform coefficient fuzzy rule base.
Optionally, the wavelet used in the wavelet transform is a mexica-hat wavelet.
The technical scheme provided by the embodiment of the invention at least has the following technical effects or advantages: the method solves the technical problems that in the prior art, the real-time performance is poor and the diagnosis accuracy is low when the fault of the complex system is diagnosed, and achieves the technical effect of quickly and accurately identifying the tiny fault in the closed-loop control system. Firstly, establishing a virtual numerical control system based on a data assimilation technology for comparing with a real closed-loop control system, and avoiding the problem that a tiny fault cannot be identified due to the structural compensation or fault tolerance of closed-loop control in the closed-loop control system; secondly, fault parameters are extracted by adopting continuous wavelet transformation, and tiny faults are identified, so that the method has the characteristics of high efficiency and high speed; in addition, the fault parameters are segmented through the membership function, the faults are divided into a limited number of types, and the effectiveness of fault detection and diagnosis is improved; finally, the initial fault is identified based on the self-adaptive fuzzy reasoning technology, the technical problem of misjudgment possibly caused by the improper rigid threshold setting method is avoided, and the identified fault is more in line with the actual field requirement.
Drawings
In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the embodiments or the prior art will be briefly described below, it is obvious that the drawings in the following description are only embodiments of the present invention, and for those skilled in the art, other drawings can be obtained according to the provided drawings without creative efforts.
FIG. 1 is a flowchart of a method for early fault detection and diagnosis of a closed-loop control system according to an embodiment of the present disclosure;
fig. 2 is a schematic structural diagram of a real closed-loop control system and a virtual closed-loop control system provided in an embodiment of the present application;
fig. 3 is a flowchart of step S2 according to a first embodiment of the present application;
fig. 4 is a flowchart of step S3 according to a first embodiment of the present application;
fig. 5 is a flowchart of step S32 according to a first embodiment of the present application;
FIG. 6 is a schematic diagram of an early fault detection and diagnosis system of a closed-loop control system according to a second embodiment of the present application;
fig. 7 is a schematic diagram of a residual error analysis and acquisition module of an early fault detection and diagnosis system of a closed-loop control system according to a second embodiment of the present disclosure;
FIG. 8 is a schematic diagram of an early fault detection and diagnosis system fault identification module of a closed-loop control system according to a second embodiment of the present disclosure;
FIG. 9 is a schematic diagram of a fuzzy rule base building module of an early failure detection and diagnosis system of a closed-loop control system according to a second embodiment of the present disclosure;
Detailed Description
The invention provides a method and a system for detecting and diagnosing early faults of a closed-loop control system, which solve the technical problems that in the prior art, when the faults of a complex system are diagnosed, the real-time performance is poor, the diagnosis accuracy rate is low, and the detection efficiency is not high, and realize the purpose of quickly and accurately identifying tiny faults in the closed-loop control system. The method comprises the steps of firstly establishing a virtual closed-loop control system based on a data assimilation technology, and comparing the virtual closed-loop control system with a real closed-loop control system, so that the problem that tiny faults cannot be identified due to structural compensation or fault tolerance of closed-loop control in the closed-loop control system can be avoided; secondly, fault parameters are extracted by adopting continuous wavelet transformation, and tiny faults are identified through the fault parameters, so that the method has the characteristics of high efficiency and high speed; in addition, the fault parameters are segmented through the membership function, the faults are divided into a limited number of types, and the effectiveness of fault detection and diagnosis is improved; finally, the initial fault is identified based on the self-adaptive fuzzy reasoning technology, the technical problem of misjudgment possibly caused by the improper rigid threshold setting method is avoided, and the identified fault is more in line with the actual field requirement.
In order to better understand the technical solution, the technical solution will be further described in detail with reference to the drawings and the specific embodiments.
Example one
An embodiment of the present application provides a method for detecting and diagnosing an early failure of a closed-loop control system, please refer to fig. 1, where the method includes the steps of:
s0, establishing a real on-site closed-loop control system according to requirements; if a real closed-loop control system is established according to an actual controlled object, measuring equipment, different controllers and the like;
s1, establishing a virtual closed-loop control system based on a data assimilation technology and a mathematical mechanism model of the equipment; establishing a virtual closed-loop control system based on a data assimilation technology, wherein the virtual closed-loop control system and a real closed-loop control system are mapped one by one; the established virtual closed-loop control system can change along with the change of the real closed-loop system without repeated reestablishment; interference caused by artificial establishment of a virtual closed-loop control system is avoided, and the accuracy of fault identification is improved;
s2, analyzing and obtaining a residual error of the real closed-loop control system relative to the virtual closed-loop control system based on the real closed-loop control system and the virtual closed-loop control system; differences caused by tiny faults can be visually seen through a numerical model; the problem that micro faults cannot be identified due to structural compensation or fault tolerance of closed-loop control is solved;
and S3, identifying the fault type in the real closed-loop control system according to the obtained residual error.
With reference to fig. 2, it can be seen that: the real closed-loop control system comprises a system input R(s), and a real closed-loop control system controller model CP(ii) a Measuring device or sensor S of a true closed-loop control systemP(ii) a Mechanism model P of controlled object of real closed-loop control systemP(ii) a Output Y of a true closed loop control systemP(s) and the regulating variable E emitted by the measuring device or sensor of the true closed-loop control systemP(s)；
The virtual closed-loop control system comprises the same system input R(s) as that in the real closed-loop control system, and a controller model C of the virtual closed-loop control systemS(ii) a Measuring device or sensor S of a virtual closed-loop control systemS(ii) a Mechanism model P of controlled object of virtual closed-loop control systemS(ii) a Output ys(s) of the virtual closed-loop control system; the real closed-loop control system and the virtual closed-loop control system are subjected to expected system disturbance D(s), except that a system minor fault F(s) exists in the real closed-loop control system.
The specific process of fault detection is as follows: in a true closed-loop control system, the true closed-loop control system inputs R(s) to a true closed-loop control system controller model CPUnder the influence of the system minor fault F(s) and the expected system disturbance D(s), the model applied to the controlled object is PPAnd finally Y is outputP(s) in which a model P of the controlled objectPThe output end is also connected with a measuring device or a sensor S of a real closed-loop control systempFor feedback in particular on the basis of measuring the output value of the true closed-loop control systemThe adjusting process is as follows: under the expected input value R(s), the control system runs smoothly, and after the control system has minor faults F(s) and disturbances D(s), the control system is arranged on a controlled object PPDownstream measuring devices or sensors SPPerforming on-line detection, and comparing the detected output value with expected input value R(s) to obtain EP(s), before reaching a disturbance and interference action point through the controller (forming a complete loop), setting a comparison point for comparing with a comparison point of a virtual control system, and screening micro fault quantity;
in a virtual closed-loop control system, the virtual closed-loop control system inputs R(s) to a controller model CSUnder the influence of the expected system disturbance D(s), the model applied to the controlled object is PSAnd finally Y is outputS(S) wherein a measuring device or sensor S of a virtual closed-loop control system is further connected to the output end of the model Ps of the controlled objectSFor measuring devices or sensors SSPerforming feedback adjustment on a real closed-loop control system, controlling the system to run stably under the same expected input value R(s) as the real control system, and setting the control system on a controlled object P after the control system generates disturbance D(s)PDownstream measuring devices or sensors SpPerforming on-line detection, and comparing the detected output value with expected input value R(s) to obtain EP(s), before reaching the disturbance and interference action point through the controller (forming a complete loop), setting a comparison point for comparing with a comparison point of a real control system, wherein the difference between the detection value of the real system and the detection value of the virtual system is a tiny fault;
from the above analysis, it can be known that the real closed-loop control system is affected by the system minor faults f(s) and the expected system disturbances d(s), and the virtual closed-loop control system is only affected by the expected system disturbances d(s), so that the residual error U of the real closed-loop control system relative to the virtual closed-loop control system can be obtained by analyzing the deviation values of the real closed-loop control system and the virtual closed-loop control system caused by the system disturbances d(s) and the minor faults f(s)F(s)；
In conjunction with fig. 3 on the basis of fig. 2, it can be seen that: step S2 includes:
s21, analyzing and obtaining a first detection deviation value U in the field real closed-loop control system according to the real closed-loop control systemP(s); the first deviation value includes a system disturbance and a minor fault signal, where UPThe expression of(s) is:
s22, analyzing and obtaining a second detection deviation value U in the virtual closed-loop control system according to the virtual closed-loop control systemS(s); the second deviation value includes only system disturbances, where USThe expression of(s) is:
and S23, acquiring a residual error of the real closed-loop control system relative to the virtual numerical control closed-loop system according to the first detection deviation value and the second detection deviation value. Because the virtual closed-loop control system is established based on a data assimilation technology and is used for diagnosing a tiny fault, the controller model, the measuring equipment or the mechanism model of the sensor and the controlled object in the virtual closed-loop control system and the real closed-loop control system are the same, namely: cP＝CS，SP＝SS，PP＝PSOn this basis, the residual U can be obtainedF(s) and UFThe expression of(s) is:
through the analysis process of the residual errors, mathematical models between the residual errors and system input, a sensor model, a controller model and the like are established, and the influence caused by the tiny faults can be visually seen through numerical values.
Further, with reference to fig. 4, step S3 includes:
s31, carrying out multi-scale and multi-resolution refinement analysis on the obtained residual error by utilizing wavelet transform, and obtaining a continuous wavelet transform coefficient of the residual error through discretization; the multi-scale and multi-resolution detailed analysis is a theory established on the concept of function space, and the basic idea is that the target can be observed from coarse to fine on each scale along with the change of the scale from large to small. The residual error is liked to be analyzed through multi-scale and multi-resolution, the characteristic points of the residual error can be comprehensively and finely obtained, further, the redundancy can be eliminated and reduced to the greatest extent through discretization under the condition that the original signal is not lost, and the follow-up processing is facilitated.
S32, constructing a residual continuous wavelet transform coefficient fuzzy rule base according to the residual continuous wavelet transform coefficients; wherein, in one embodiment of the invention, a residual continuous wavelet transform coefficient fuzzy rule base is constructed by a K-means clustering algorithm;
s33, acquiring a fault type in a real closed-loop control system according to the residual continuous wavelet transform coefficient and the residual continuous wavelet transform coefficient fuzzy rule base; the specific identification process is as follows: and inputting the residual continuous wavelet transform coefficient into the residual continuous wavelet transform coefficient fuzzy rule base for fault identification, and identifying the fault type in the real closed-loop control system. The initial fault is identified based on the fuzzy rule base, the misjudgment caused by the improper rigid threshold setting method is avoided, and the actual field requirements are better met.
Further, with reference to fig. 5, a specific process for constructing the wavelet transform coefficient fuzzy rule base is as follows:
s321, calculating the average value of the residual continuous wavelet transform coefficients, and taking the average value as a threshold; a reference value is provided, so that the tiny fault type can be better measured;
s322, calculating the sum of the residual continuous wavelet transform coefficients, calculating the difference value of the threshold value and the sum of the residual continuous wavelet transform coefficients, and dividing the sum of the residual continuous wavelet transform coefficients into a limited range according to the difference value and through a membership function; because the residual continuous wavelet transform coefficients have slight differences for different controlled objects, if fault type setting is carried out on each difference, the workload is complex and is not necessary, the sum of the residual continuous wavelet transform coefficients is divided into a limited range according to the difference through a membership function, the workload can be reduced, and the working efficiency is improved.
S323, according to the limited range of the sum of the residual continuous wavelet transform coefficients, correspondingly establishing limited fault types, and constructing a residual continuous wavelet transform coefficient fuzzy rule base. Namely: if the sum of the magnitudes of the residual continuous wavelet transform coefficients is a, then the corresponding fault is B, and in one of the examples of the present invention, low, medium, and high respectively correspond to a zero fault, a single fault, or multiple faults.
Further, in one embodiment of the present invention, the wavelet used in the wavelet transform is a mexica-hat wavelet, and the mexica-hat wavelet is more suitable for processing signals than other wavelets, and the expression is as follows:
therein, Ψt,a,bRepresenting the squared integrable function, a being the shrinkage factor, b being the translation factor, t being the time, e being a constant, being the base of the natural logarithm.
Example two
In order to implement the method for detecting and diagnosing the early failure of the closed-loop control system based on the data assimilation technology, fig. 6 is a schematic diagram of a system for detecting and diagnosing the early failure of the closed-loop control system provided by a second embodiment of the present invention, and the system includes:
a real closed-loop control system establishing module 100, configured to establish a field real closed-loop control system according to actual requirements; if a real closed-loop control system is established according to an actual controlled object, measuring equipment, different controllers and the like;
the virtual closed-loop control system establishing module 200 is connected with the real closed-loop control system establishing module 100 and is used for establishing a virtual closed-loop control system based on a data assimilation technology and a mathematical mechanism model of equipment; establishing a virtual closed-loop control system based on a data assimilation technology, wherein the virtual closed-loop control system and a real closed-loop control system are mapped one by one; the established virtual closed-loop control system can change along with the change of the real closed-loop system without repeated reestablishment; interference caused by artificial establishment of a virtual closed-loop control system is avoided, and the accuracy of fault identification is improved;
a residual error analyzing and acquiring module 300, connected to the real closed-loop control system establishing module 100 and the virtual closed-loop control system establishing module 200, for analyzing and acquiring a residual error of the real closed-loop control system relative to the virtual numerical control system based on the real closed-loop control system and the virtual closed-loop control system; differences caused by tiny faults can be visually seen through a numerical model; the problem that micro faults cannot be identified due to structural compensation or fault tolerance of closed-loop control is solved;
and the fault identification module 400 is connected with the residual error analysis and acquisition module 300 and is used for identifying the fault type in the real closed-loop control system according to the obtained residual error.
Further, referring to fig. 7, the residual analysis and acquisition module 300 includes:
the real closed-loop control system is influenced by system minor faults F(s) and expected system disturbances D(s), and the virtual closed-loop control system is only influenced by the expected system disturbances D(s), so that the residual error U of the real closed-loop control system relative to the virtual closed-loop control system can be obtained by analyzing deviation values of the real closed-loop control system and the virtual closed-loop control system caused by the system disturbances D(s) and the minor faults F(s)F(s); the method specifically comprises the following steps:
a first detection deviation value obtaining module 310, connected to the real closed-loop control system establishing module 100, for analyzing and obtaining a first detection deviation value U in the on-site real closed-loop control system according to the real closed-loop control systemP(s); the first deviation value comprises a system deviation and a minor fault signal, wherein UPThe expression of(s) is:
a second detection deviation value obtaining module 320 connected to the virtual numerical closed-loop control system establishing module 200 for analyzing and obtaining a second detection deviation value U in the virtual closed-loop control systemS(s); the second deviation value includes only systematic deviations, where USThe expression of(s) is:
a residual error obtaining module 330, connected to the first detected deviation value obtaining module 310 and the second detected deviation value obtaining module 320, for obtaining a residual error U of the real closed-loop control system relative to the virtual closed-loop control closed-loop system according to the first detected deviation value and the second detected deviation valueF(s) and UFThe expression of(s) is:
further, with reference to fig. 8, the fault identification module 400 includes:
a continuous wavelet transform coefficient obtaining module 410, connected to the residual error obtaining module 330, configured to perform multi-scale multi-resolution refinement analysis on the obtained residual error by using wavelet transform, and obtain a continuous wavelet transform coefficient of the residual error through discretization; the residual error is liked to be analyzed through multi-scale and multi-resolution, the characteristic points of the residual error can be comprehensively and finely obtained, further, the redundancy can be eliminated and reduced to the greatest extent through discretization under the condition that the original signal is not lost, and the follow-up processing is facilitated.
A fuzzy rule base establishing module 420 connected to the continuous wavelet transform coefficient obtaining module 410, for establishing a residual continuous wavelet transform coefficient fuzzy rule base according to the residual continuous wavelet transform coefficients;
and the fault type identification module 430 is connected with the continuous wavelet transform coefficient acquisition module 410 and the fuzzy rule base establishment module 420, and is configured to acquire a fault type in the real closed-loop control system according to the residual continuous wavelet transform coefficient fuzzy rule base. The initial fault is identified based on the fuzzy rule base, the misjudgment caused by the improper rigid threshold setting method is avoided, and the actual field requirements are better met.
Further, in conjunction with fig. 9, the fuzzy rule base building module 420 includes:
a threshold module 421 connected to the continuous wavelet transform coefficient obtaining module 410, configured to calculate an average value of residual continuous wavelet transform coefficients, and use the average value as a threshold; the establishment of the threshold value provides a reference for measuring the minor fault.
A segmentation module 422, connected to the continuous wavelet transform coefficient acquisition module 410 and the threshold module 421, for calculating the sum of residual continuous wavelet transform coefficients, calculating the difference between the threshold and the sum of residual continuous wavelet transform coefficients, and dividing the sum of residual continuous wavelet transform coefficients into mexica-hat wavelet ranges according to the difference and by means of membership functions; because the residual continuous wavelet transform coefficients have slight differences for different controlled objects, if fault type setting is carried out on each difference, the workload is complex and is unnecessary, in one embodiment of the invention, the sum of the residual continuous wavelet transform coefficients is divided into three ranges according to the difference through a membership function, and the three ranges are respectively low, medium and high, so that the workload can be reduced while the fault identification is ensured to be accurate to a great extent, and the working efficiency is improved.
The fuzzy rule base rule establishing module 423 is connected with the continuous wavelet transform coefficient obtaining module 410 and the segmenting module 422, and is used for correspondingly establishing limited fault types according to the limited range of the sum of the residual continuous wavelet transform coefficients and establishing a residual continuous wavelet transform coefficient fuzzy rule base. In one example of the present invention, low, medium, and high correspond to a zero fault, a single fault, or multiple faults, respectively.
Wherein, the wavelet adopted by the wavelet transform is mexica-hat wavelet.
The method and the system for detecting and diagnosing the early fault of the closed-loop control system based on the data assimilation technology have the following technical effects: establishing a virtual closed-loop control system based on a data assimilation technology, wherein the virtual closed-loop control system and a real closed-loop control system are mapped one by one; the established virtual closed-loop control system can change along with the change of the real closed-loop system without repeated reestablishment; interference caused by artificial establishment of a virtual closed-loop control system is avoided, and the accuracy of fault identification is improved; meanwhile, the invention judges the minor fault by establishing mathematical models between the residual error and the system input, the sensor precision, the controller model and the like, and the difference caused by the minor fault can be visually seen through the numerical model, so that the influence of each parameter on the residual error can be quantized; the invention extracts the fault parameters through continuous wavelet transformation, namely mexica-hat wavelet, can quickly and effectively identify the tiny fault and improve the working efficiency; in addition, the sum of residual continuous wavelet transform coefficients is divided into three ranges through a membership function; the invention is based on the self-adaptive fuzzy reasoning technology, namely, the fault is identified by establishing the fuzzy rule base, so that the misjudgment caused by the improper rigid threshold setting method is avoided, and the actual field requirements are better met.
It should be noted that: in the embodiment, when the training method is implemented, the system is only illustrated by dividing the functional modules, and in practical application, the functions may be distributed by different functional modules according to needs, that is, the internal structure of the device is divided into different functional modules, so as to complete all or part of the functions described above. In addition, the system and method embodiments provided by the above embodiments belong to the same concept, and specific implementation processes thereof are described in detail in the method embodiments and are not described herein again.
Those skilled in the art will appreciate that all or part of the steps for implementing the above embodiments may be implemented by hardware, or may be implemented by a program instructing associated hardware, and the program may be stored in a computer readable storage medium.
The above description is only for the purpose of illustrating the preferred embodiments of the present invention and is not to be construed as limiting the invention, and any modifications, equivalents, improvements and the like that fall within the spirit and principle of the present invention are intended to be included therein.
Claims (8)
1. A closed loop control system early fault detection and diagnosis method, the method comprising:
establishing a field real closed-loop control system according to actual requirements;
establishing a virtual closed-loop control system based on a data assimilation technology and a mathematical mechanism model of equipment;
analyzing and obtaining a residual error of the real closed-loop control system relative to the virtual closed-loop control system based on the real closed-loop control system and the virtual closed-loop control system;
carrying out multi-scale and multi-resolution refinement analysis on the obtained residual errors by utilizing wavelet transformation, and obtaining residual error continuous wavelet transformation coefficients through discretization; constructing a residual continuous wavelet transform coefficient fuzzy rule base according to the residual continuous wavelet transform coefficients; and identifying the fault type in the real closed-loop control system according to the residual continuous wavelet transform coefficient and the residual continuous wavelet transform coefficient fuzzy rule base.
2. The early fault detection and diagnosis method of claim 1, wherein said obtaining a residual of said real closed-loop-control system relative to said virtual closed-loop-control system based on said real closed-loop-control system and virtual closed-loop-control system analysis further comprises:
analyzing and obtaining a first detection deviation value in the field real closed-loop control system according to the real closed-loop control system;
analyzing and obtaining a second detection deviation value in the virtual closed-loop control system according to the virtual closed-loop control system;
and acquiring a residual error of the real closed-loop control system relative to the virtual closed-loop control system according to the first detection deviation value and the second detection deviation value.
3. The early fault detection and diagnosis method of claim 1, wherein said constructing a residual continuous wavelet transform coefficient fuzzy rule base from said residual continuous wavelet transform coefficients further comprises:
calculating the average value of the residual continuous wavelet transform coefficient, and taking the average value as a threshold value;
calculating the sum of the residual continuous wavelet transform coefficients, calculating the difference value of the threshold value and the sum of the residual continuous wavelet transform coefficients, and dividing the sum of the residual continuous wavelet transform coefficients into a limited range according to the difference value and through a membership function;
and correspondingly establishing limited fault types according to the limited range of the sum of the residual continuous wavelet transform coefficients, and establishing a residual continuous wavelet transform coefficient fuzzy rule base.
4. The early fault detection and diagnosis method of claim 1, wherein the wavelet used in the wavelet transform is a mexica-hat wavelet.
5. A closed loop control system early fault detection and diagnosis system, the system comprising:
the real closed-loop control system establishing module is used for establishing a field real closed-loop control system according to requirements;
the virtual closed-loop control system establishing module is connected with the real closed-loop control system establishing module and is used for establishing a virtual closed-loop control system based on a data assimilation technology and a mathematical mechanism model of equipment;
the residual error analysis and acquisition module is connected with the real closed-loop control system and the virtual closed-loop control system establishment module and is used for analyzing and acquiring the residual error of the real closed-loop control system relative to the virtual closed-loop control system based on the real closed-loop control system and the virtual closed-loop control system;
a fault identification module, the fault identification module comprising:
the continuous wavelet transform coefficient acquisition module is connected with the residual error analysis and acquisition module and is used for carrying out multi-scale and multi-resolution refinement analysis on the obtained residual error by utilizing wavelet transform and obtaining a residual error continuous wavelet transform coefficient through discretization;
the fuzzy rule base establishing module is connected with the continuous wavelet transform coefficient acquiring module and used for establishing a residual continuous wavelet transform coefficient fuzzy rule base according to the residual continuous wavelet transform coefficients;
and the fault type identification module is connected with the continuous wavelet transform coefficient acquisition module and the fuzzy rule base establishment module and is used for acquiring the fault type in the real closed-loop control system according to the residual continuous wavelet transform coefficient and the residual continuous wavelet transform coefficient fuzzy rule base.
6. The early fault detection and diagnosis system of claim 5, wherein the residual analysis and acquisition module comprises:
the first detection deviation value acquisition module is connected with the real closed-loop control system establishment module and used for analyzing and acquiring a first detection deviation value in the field real closed-loop control system according to the real closed-loop control system;
the second detection deviation value acquisition module is connected with the virtual closed-loop control system establishment module and used for analyzing and acquiring a second detection deviation value in the virtual closed-loop control system;
and the residual error acquisition module is connected with the first detection deviation value acquisition module and the second detection deviation value acquisition module and is used for acquiring the residual error of the real closed-loop control system relative to the virtual closed-loop control system according to the first detection deviation value and the second detection deviation value.
7. The early fault detection and diagnosis system of claim 5, wherein the fuzzy rule base establishment module comprises:
the threshold module is connected with the continuous wavelet transform coefficient acquisition module and used for calculating the average value of the residual continuous wavelet transform coefficients and taking the average value as a threshold;
the segmentation module is connected with the continuous wavelet transform coefficient acquisition module and the threshold module and used for calculating the sum of the residual continuous wavelet transform coefficients, calculating the difference value of the threshold and the sum of the residual continuous wavelet transform coefficients, and dividing the sum of the residual continuous wavelet transform coefficients into a limited range according to the difference value and through a membership function;
and the fuzzy rule base rule establishing module is connected with the continuous wavelet transform coefficient acquisition module and the segmentation module and is used for correspondingly establishing limited fault types according to the limited range of the sum of the residual continuous wavelet transform coefficients and establishing a residual continuous wavelet transform coefficient fuzzy rule base.
8. The early fault detection and diagnosis system of claim 5, wherein the wavelet used in said wavelet transform is a mexica-hat wavelet.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910312767.0A CN110045716B (en) | 2019-04-18 | 2019-04-18 | Method and system for detecting and diagnosing early fault of closed-loop control system |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910312767.0A CN110045716B (en) | 2019-04-18 | 2019-04-18 | Method and system for detecting and diagnosing early fault of closed-loop control system |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN110045716A CN110045716A (en) | 2019-07-23 |
| CN110045716B true CN110045716B (en) | 2020-08-11 |
Family
ID=67277791
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910312767.0A Active CN110045716B (en) | 2019-04-18 | 2019-04-18 | Method and system for detecting and diagnosing early fault of closed-loop control system |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110045716B (en) |
Families Citing this family (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112906473B (en) * | 2021-01-19 | 2023-06-20 | 杭州安脉盛智能技术有限公司 | Fault diagnosis method for rotary equipment |
Family Cites Families (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US6996442B2 (en) * | 2003-01-15 | 2006-02-07 | Xerox Corporation | Systems and methods for detecting impending faults within closed-loop control systems |
| CN103941725B (en) * | 2014-04-24 | 2017-09-08 | 淮海工学院 | A kind of method for diagnosing faults of nonlinear networked control systems |
| CN105807611B (en) * | 2016-03-05 | 2018-06-12 | 华中科技大学 | A kind of model of closed-loop control system and the unmatched detection method of object |
| CN107272660B (en) * | 2017-07-26 | 2019-05-17 | 江南大学 | A kind of random fault detection method of the network control system with packet loss |
| CN108667673B (en) * | 2018-06-22 | 2022-02-22 | 东北大学 | Nonlinear network control system fault detection method based on event trigger mechanism |
| CN109542080B (en) * | 2018-10-16 | 2020-05-08 | 中国人民解放军国防科技大学 | Closed-loop control system fault detection method based on dual space transformation |
-
2019
- 2019-04-18 CN CN201910312767.0A patent/CN110045716B/en active Active
Also Published As
| Publication number | Publication date |
|---|---|
| CN110045716A (en) | 2019-07-23 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Zhang et al. | LSTM-based analysis of industrial IoT equipment | |
| CN107941537B (en) | A kind of mechanical equipment health state evaluation method | |
| Khelif et al. | Direct remaining useful life estimation based on support vector regression | |
| Bezerra et al. | An evolving approach to unsupervised and real-time fault detection in industrial processes | |
| Lindemann et al. | Anomaly detection and prediction in discrete manufacturing based on cooperative LSTM networks | |
| KR102189269B1 (en) | Fault Diagnosis method and system for induction motor using convolutional neural network | |
| CN109146246B (en) | Fault detection method based on automatic encoder and Bayesian network | |
| CN113642754B (en) | Complex industrial process fault prediction method based on RF noise reduction self-coding information reconstruction and time convolution network | |
| Amiruddin et al. | Valve stiction detection through improved pattern recognition using neural networks | |
| CN112284440B (en) | Sensor data deviation self-adaptive correction method | |
| Baghernezhad et al. | Computationally intelligent strategies for robust fault detection, isolation, and identification of mobile robots | |
| Wei et al. | A novel deep learning model based on target transformer for fault diagnosis of chemical process | |
| Mathew et al. | Regression kernel for prognostics with support vector machines | |
| Wang et al. | Adaptive prognosis of centrifugal pump under variable operating conditions | |
| CN110045716B (en) | Method and system for detecting and diagnosing early fault of closed-loop control system | |
| CN109324591B (en) | fault diagnosis device and method of stochastic hybrid system based on robust estimator | |
| CN113158448A (en) | Self-adaptive calculation method for equipment threshold of ship system | |
| Aremu et al. | Kullback-leibler divergence constructed health indicator for data-driven predictive maintenance of multi-sensor systems | |
| CN116842379A (en) | Mechanical bearing residual service life prediction method based on DRSN-CS and BiGRU+MLP models | |
| Kakati et al. | Remaining useful life predictions for turbofan engine degradation using online long short-term memory network | |
| Kourd et al. | Early FDI based on residuals design according to the analysis of models of faults: application to DAMADICS | |
| Wang et al. | Nonlinear system identification and fault detection using hierarchical clustering analysis and local linear models | |
| CN114037012A (en) | Flight data anomaly detection method based on correlation analysis and deep learning | |
| Zhao et al. | A unified framework for fault detection and diagnosis using particle filter | |
| Shimizu et al. | Real-time techniques for fault detection on railway door systems |
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 |