A model-based fault diagnosis method for measurement and control equipment

Abstract

The present invention discloses a fault diagnosis method for measurement and control equipment based on a model, which specifically includes the following steps: S1, model definition; S2, logic definition; S3, directed graph conversion; S4, traversal positioning: The present invention relates to the field of aerospace measurement and control technology. The fault diagnosis method for measurement and control equipment based on a model uses processing functions and observation functions as basic mathematical models for single node fault diagnosis, establishes all the mutual relations between input and output signals of various functional modules inside the measurement and control equipment system, realizes the description of all possible states inside the measurement and control equipment system, can cover all possible faults of the measurement and control equipment system, and adopts a method for describing the mutual relations between various functional modules of the measurement and control equipment system with a directed graph, realizes rapid diagnosis and precise positioning of functional module faults through directed graph search, and the diagnosis model is relatively simple, the combination process is not complicated, and has good scalability.

Description

Technical Field

[0001] The invention relates to the field of aerospace measurement and control technology, and is mainly applied to intelligent management of aerospace measurement and control equipment, specifically a model-based measurement and control equipment fault diagnosis method. Background Art

[0002] Traditional measurement and control equipment fault diagnosis is basically equivalent to abnormal alarm, that is, the equipment status is monitored through the monitoring points set in the system to understand the working conditions of the relevant parts of the equipment, and alarm is issued for the status that exceeds the normal value range. This fault detection method has the following problems:

[0003] 1. The probability of false reporting, missed reporting and wrong reporting of faults is high: it can only make simple judgments based on the status of the monitoring points, the fault judgment conditions are single, it cannot adapt to various complex operating conditions, and the misdiagnosis rate is high;

[0004] 2. Unable to accurately locate the faulty unit: After an abnormal alarm occurs, it is impossible to accurately locate the fault point, and it is impossible to give the specific cause of the fault and solution, and it is impossible to guide the operator to repair the fault.

[0005] To solve the above problems, the more common measurement and control equipment fault diagnosis system at present is the fault diagnosis expert system. Its principle is to collect and standardize the knowledge of industry experts to form rules that can be used for automatic reasoning, and then the knowledge reasoning engine reasons according to the rules to analyze the fault location. The accuracy of fault diagnosis using this scheme depends entirely on the comprehensiveness of expert knowledge. That is, in order to achieve accurate fault diagnosis, the prerequisite is that the knowledge of all possible faults of the target system must be collected, that is, exhaustive. This kind of knowledge is a combinatorial knowledge. For complex systems, there will inevitably be a combinatorial explosion of knowledge, which is actually impossible to exhaust. It is the existence of this knowledge combinatorial explosion phenomenon that makes it difficult to further improve the diagnostic accuracy of the fault diagnosis expert system in actual engineering applications, and it is also difficult to accurately locate the location of the fault. Summary of the invention

[0006] 1. Technical issues to be solved

[0007] In view of the deficiencies in the prior art, the present invention provides a model-based measurement and control equipment fault diagnosis method, which solves the problem that in actual engineering applications, the diagnostic accuracy of the fault diagnosis expert system is difficult to further improve and the location of the fault is difficult to accurately locate.

[0008] (II) Technical solution

[0009] To achieve the above object, the present invention provides the following technical solution: a model-based fault diagnosis method for measurement and control equipment, specifically comprising the following steps:

[0010] S1. Model definition: Several functional modules in the measurement and control equipment system are defined as several nodes. After the node processes a specific input signal x, the corresponding output signal y is obtained. The working function of the node is represented by f. The process can be represented by: y = f(x), that is, the output signal y is the result of the input signal x after being transformed by the function f. The function f is defined as the processing function of the node. For any node, there are several monitoring parameters S, which are used to monitor the node status. As a certain function processing result of the working condition of the node, this processing function is represented by G. The process can be represented by: S = G(f), that is, the monitoring parameter S is the result of the processing function f after being transformed by the function G. The function G is defined as the observation function of the node.

[0011] S2. Logical definition: define the logical function of the processing function f and the observation function G, define the normal state as 1, and the abnormal state as 0. For the processing function f, if its input parameter x1 is 1, then f(x1, x2, ...) = 1; otherwise, f(x1, x2, ...) = 0. For the observation function G, if f is normal, then G(f) = (s1, s2, ...), where s i The value of is 1, indicating that the node is fault-free. If f is abnormal, then G(f)=(s1', s2', ...), where at least one s i ' is 0, indicating that one of the nodes is faulty;

[0012] S3, directed graph conversion: Arrange the nodes in the directed graph corresponding to the nodes of the measurement and control equipment system, connect the nodes in the directed graph corresponding to the flow direction of the signals of various nodes, define the processing function f and observation function G of the node to represent the properties of the node, and find the nodes in the directed graph where the processing function f works abnormally according to the input signal x, output signal y, and the value of the monitoring parameter S;

[0013] S4, traversal and positioning: After completing the conversion of the directed graph in S3, starting from the input node, for each node in the directed graph, calculate its output signal y and the theoretical value of the observation parameter S according to the value of all the input signals x of the node, and compare the calculated theoretical value with the actual value obtained by actual observation. If there is a difference, it indicates that a fault has occurred at this node, and fault diagnosis and positioning are achieved. Otherwise, continue to calculate the next node until the traversal calculation of all nodes is completed.

[0014] By adopting the above technical scheme, the processing function and the observation function are used as the basic mathematical model for single node fault diagnosis, and all the relationships between the input and output signals of the functional modules inside the measurement and control equipment system are established, so as to realize the description of all possible states inside the measurement and control equipment system, and cover all possible faults of the measurement and control equipment system. In addition, a method of using a directed graph to describe the relationship between the functional modules of the measurement and control equipment system is adopted, and through directed graph search, rapid diagnosis and precise positioning of functional module faults can be realized.

[0015] The present invention is further configured as follows: when searching for abnormal nodes in S4, a processing function f of each node in the measurement and control equipment system is given i , observation function G i , and given the external input signal x of the measurement and control equipment system j , the measurement and control equipment system outputs signals to the outside j , and monitoring parameters of the measurement and control equipment system k .

[0016] The present invention is further configured as follows: in S4, a theoretical value of the output signal y is calculated by processing function f.

[0017] The present invention is further configured as follows: in S4, the theoretical value of the monitoring parameter S is calculated by the observation function G.

[0018] The present invention is further configured as follows: since the intermediate result is not available, the location of the faulty node cannot be determined only by the processing function f, and the observation function G is introduced for joint determination.

[0019] The present invention is further configured as follows: the state monitoring parameters of any node are divided into two categories, one category is the parameters that directly describe the state of the node itself, and the parameters are represented by the observation function G; the other category is the parameters that describe the input signal of the node, and the parameters correspond to the output variables in the signal transmission process of any node. If the state is normal, it indicates that all links of the signal from the input to the node are normal; otherwise, it indicates that there is a fault point in the signal transmission process.

[0020] The present invention is further configured as follows: when searching for a node in the directed graph where the processing function f works abnormally, a credibility analysis is performed on the input signal x, the output signal y, and the value of the monitoring parameter S, and a contribution rate D of the values ​​of x, y, and S at a certain node is calculated. The greater the proportion of the contribution rate, the higher the probability that the node causes equipment failure. The calculation steps of the contribution rate D are as follows:

[0021] S31: Collect node samples of input signal x, output signal y and monitoring parameter S;

[0022] S32: Calculate the mean of the above node samples and

[0023] S33: Calculate the characteristic value λx of each node sample i , λy i and λS i ;

[0024] S34: Calculate the contribution rate D of the x, y, and S values ​​at a certain node.

[0025] The present invention is further configured as follows: the value of the input signal x in samples of all nodes can be expressed as x=[x1, x2, x3, ..., x n ], the value of the output signal y in all node samples can be expressed as y = [y1, y2, y3, ..., y n ], the value of the monitoring parameter S in all node samples can be expressed as S = [S1, S2, S3, ..., S n ]; then the mean of the input signal x is The mean of the output signal y is The mean value of the monitoring parameter S is Eigenvalue Eigenvalue Eigenvalue The contribution rates D of x, y, and S values ​​at a certain node are respectively Where i=1,2,3,…,n.

[0026] The present invention is further configured as follows: when D i The larger the value, the higher the probability that the node causes equipment failure and the higher the credibility. When troubleshooting, equipment problems at this node can be prioritized.

[0027] (III) Beneficial effects

[0028] The present invention provides a model-based fault diagnosis method for measurement and control equipment, which has the following beneficial effects:

[0029] (1) This model-based measurement and control equipment fault diagnosis method uses the processing function and the observation function as the basic mathematical model for single node fault diagnosis, establishes all the relationships between the input and output signals of each functional module within the measurement and control equipment system, and realizes the description of all possible states within the measurement and control equipment system. It can cover all possible faults of the measurement and control equipment system, and adopts a directed graph to describe the relationship between the functional modules of the measurement and control equipment system. Through directed graph search, the rapid diagnosis and precise positioning of the functional module faults can be achieved.

[0030] (2) The model-based fault diagnosis method for measurement and control equipment first establishes the fault diagnosis model of the functional module and then combines these basic functional diagnosis models to form a fault model of the measurement and control equipment system. Since the fault diagnosis model of a single functional module is relatively simple and the model combination process is not complicated, the construction of the entire measurement and control equipment system model is relatively simple. Moreover, the models of several measurement and control equipment systems can be used as components and further combined into a larger measurement and control equipment system fault diagnosis model, thereby constructing a fault diagnosis model for larger-scale and more complex systems, which has good scalability. BRIEF DESCRIPTION OF THE DRAWINGS

[0031] Figure 1 is a directed graph of the present invention;

[0032] Figure 2 This is a schematic diagram of the process flow of Comparative Example 1 of the present invention;

[0033] Figure 3 This is a principle block diagram of a typical measurement and control device in the present invention. DETAILED DESCRIPTION

[0034] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0035] See also Figure 1-3 , the present invention provides the following technical solutions:

[0036] Embodiment 1

[0037] A model-based fault diagnosis method for measurement and control equipment, specifically comprising the following steps:

[0038] S1. Model definition: Several functional modules in the measurement and control equipment system are defined as several nodes. After the node processes a specific input signal x, the corresponding output signal y is obtained. The working function of the node is represented by f. The process can be represented by: y = f(x), that is, the output signal y is the result of the input signal x after being transformed by the function f. The function f is defined as the processing function of the node. For any node, there are several monitoring parameters S, which are used to monitor the node status. As a certain function processing result of the working condition of the node, this processing function is represented by G. The process can be represented by: S = G(f), that is, the monitoring parameter S is the result of the processing function f after being transformed by the function G. The function G is defined as the observation function of the node.

[0039] S2. Logical definition: define the logical function of the processing function f and the observation function G, define the normal state as 1, and the abnormal state as 0. For the processing function f, if its input parameter x1 is 1, then f(x1, x2, ...) = 1; otherwise, f(x1, x2, ...) = 0. For the observation function G, if f is normal, then G(f) = (s1, s2, ...), where s i The value of is 1, indicating that the node is fault-free. If f is abnormal, then G(f)=(s1', s2', ...), where at least one s i ' is 0, indicating that one of the nodes is faulty;

[0040] S3, directed graph conversion: Arrange the nodes in the directed graph corresponding to the nodes of the measurement and control equipment system, connect the nodes in the directed graph corresponding to the flow direction of the signals of various nodes, define the processing function f and observation function G of the node to represent the properties of the node, and find the nodes in the directed graph where the processing function f works abnormally according to the input signal x, output signal y, and the value of the monitoring parameter S;

[0041] S4, traversal positioning: After completing the conversion of the directed graph in S3, the processing function f of each node in the measurement and control equipment system is given i , observation function G i , and given the external input signal x of the measurement and control equipment system j , the measurement and control equipment system outputs signals to the outside j , and monitoring parameters of the measurement and control equipment system k , starting from the input node, for each node in the directed graph, according to the value of all the input signals x of the node, the theoretical value of its output signal y is calculated by the processing function f, and the theoretical value of the monitoring parameter S is calculated by the observation function G. The calculated theoretical value is compared with the actual value obtained by actual observation. If there is a difference, it indicates that a fault has occurred at this node, and fault diagnosis and location are achieved. Otherwise, continue to calculate the next node until the traversal calculation of all nodes is completed.

[0042] As a detailed explanation, f is the work function of the node, which is specified during design and thus its form is known;

[0043] x and y are the input and output signals of the node, which are known in a specific link (input or output node of the whole system) or under specific conditions, but in most cases they are unknown;

[0044] G is the state processing function of the node. In actual fault diagnosis applications, this function G is expressed as a simple, known mapping form;

[0045] S is the monitoring parameter of the node, which is a known variable set. The specific parameter content in the set is related to the node type.

[0046] Comparative Example 1

[0047] Any set of measurement and control equipment is composed of several functional modules. The function of each module can be regarded as processing the input signal to form the corresponding output signal. The output signals of some modules are the input signals of other modules. All modules and signals are combined to complete the overall function of the measurement and control equipment. The schematic diagram of a typical measurement and control equipment is shown in the attached figure. Figure 3 As shown in the figure, taking a typical measurement and control device as an example, its input is the RF output signal generated by the antenna, and its output is a digital signal formed after the change processing. The flow direction of the signal, that is, the processing process, is a serial process. The input signal x is processed by the waveguide feeding network f1, X cryogenic receiver f2, RF matrix f3, X down-conversion f4, intermediate frequency matrix f5, signal acquisition and recording f6 and other nodes in sequence, and finally forms the output signal y. At the same time, in order to ensure its normal operation, external signals such as time and frequency are input to it. These external signals also constitute the input of the system. In the system, it also has its own flow direction, that is, processing process and related nodes, including time and frequency optical receiver f6. 11 、Pulse splitter f 12 、B code split f 13 , frequency standard phase lock f 14 , frequency mark splitter f 15 , High LO A f 16 、Low LO A f 18 , low local oscillator B f 19 , switch shunt f 20 The specific process is as follows Figure 2 As shown, therefore, for a certain input signal x, the working model can be described by the following processing function group:

[0048] y1=f1(x)

[0049] y2=f2(y1)

[0050] y3=f3(y2)

[0051] y4=f4(y3,z7,z9)

[0052] y5=f5(y4)

[0053] y6=f6(y5,v2,w2)

[0054] z1=f 11 (t)

[0055] z2=f 14 (z1)

[0056] z3=f 15 (z2)

[0057] z7=f 16 (z3)

[0058] z5=f 15 (z2)

[0059] z8=f 18 (z5)

[0060] z9=f 20 (z8)

[0061] v1=f 11 (t)

[0062] v2=f 13 (v1)

[0063] w1=f 11 (t)

[0064] w2=f 12 (w1).

[0065] Obviously, in the attached Figure 2 In the example, when the input x and t are known quantities and the system operates normally, the output y6 will be a completely certain value, that is, the expected value. When the actual measured y6 value is different from the expected value, it indicates that the system has a fault, that is, a node f i A failure occurs. However, since the intermediate results of the above process are usually unavailable, when the y6 value is found to be abnormal, the location of the node where the failure occurs cannot be determined by relying solely on the above working function group.

[0066] In summary, the present invention introduces the observation function G of each node to perform joint judgment. Specifically, the state monitoring parameters of a node are divided into two categories. One category is the parameters that directly describe the state of the node itself, which is an indication signal of node failure and can be obtained through the observation function G; the other category is the parameters that describe the input signal of the node, which actually corresponds to the output variable of an intermediate node in the transmission process of a certain signal. If the state is normal, it indicates that all processing links (constituting a local path) of this signal from the input to the previous node are normal. Otherwise, it indicates that there is a faulty node in this local path, so that the location where the fault may occur can be refined and located.

[0067] Moreover, for a complete set of measurement and control equipment, the detailed signal connection relationship between nodes and the processing function f of each node are given. Then, for any given input signal x, after several function transformations, the corresponding theoretical signal y can be obtained. Therefore, if the actual output of the equipment obtained by actual measurement is inconsistent with the theoretical output obtained by function calculation, it indicates that a node in the equipment has a fault. Furthermore, since the theoretical output of any position inside the equipment can be calculated, if the actual output of a node inside the equipment is obtained, then this actual output can be compared with the theoretical output to determine whether there is a faulty node between the starting node of the equipment and the corresponding node. When a system fault occurs, as long as there is an actual output observation point, the above steps can be repeated to eventually locate the faulty node, thereby realizing equipment fault diagnosis.

[0068] At the same time, when searching for nodes in the directed graph where the processing function f works abnormally, the credibility analysis is performed on the input signal x, the output signal y, and the monitoring parameter S values, and the contribution rate D of the x, y, and S values ​​at a certain node is calculated. The larger the contribution rate, the higher the probability that the node causes equipment failure. The failure point of the equipment is further determined. The calculation steps of the contribution rate D are:

[0069] S31: Collect node samples of input signal x, output signal y and monitoring parameter S;

[0070] S32: Calculate the mean of the above node samples and

[0071] S33: Calculate the characteristic value λx of each node sample i , λy i and λS i ;

[0072] S34: Calculate the contribution rate D of the x, y, and S values ​​at a certain node.

[0073] The value of the input signal x in all nodes can be expressed as x = [x1, x2, x3, ..., x n ], the value of the output signal y in all node samples can be expressed as y = [y1, y2, y3, ..., y n ], the value of the monitoring parameter S in all node samples can be expressed as S = [S1, S2, S3, ..., S n ]; then the mean of the input signal x is The mean of the output signal y is The mean value of the monitoring parameter S is Eigenvalue Eigenvalue Eigenvalue The contribution rates D of x, y, and S values ​​at a certain node are respectively Where i=1,2,3,…,n. When D i The larger the value, the higher the probability that the node causes equipment failure and the higher the credibility. When troubleshooting, equipment problems at this node can be prioritized.

[0074] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A model-based fault diagnosis method for measurement and control equipment, characterized in that: The specific steps include: S1. Model definition: Several functional modules in the measurement and control equipment system are defined as several nodes. After the node processes a specific input signal x, the corresponding output signal y is obtained. The working function of the node is represented by f. The process can be represented by: y = f(x), that is, the output signal y is the result of the input signal x after being transformed by the function f. The function f is defined as the processing function of the node. For any node, there are several monitoring parameters S, which are used to monitor the node status. As a certain function processing result of the working condition of the node, this processing function is represented by G. The process can be represented by: S = G(f), that is, the monitoring parameter S is the result of the processing function f after being transformed by the function G. The function G is defined as the observation function of the node. S2. Logical definition: define the logical function of the processing function f and the observation function G, define the normal state as 1, and the abnormal state as 0. For the processing function f, if its input parameter x1 is 1, then f(x1, x2, ...) = 1; otherwise, f(x1, x2, ...) = 0. For the observation function G, if f is normal, then G(f) = (s1, s2, ...), where s i The value of is 1, indicating that the node is fault-free. If f is abnormal, then G(f)=(s1', s2', ...), where at least one s i ' is 0, indicating that one of the nodes is faulty; S3, directed graph conversion: Arrange the nodes in the directed graph corresponding to the nodes of the measurement and control equipment system, connect the nodes in the directed graph corresponding to the flow direction of the signals of various nodes, define the processing function f and observation function G of the node to represent the properties of the node, and find the nodes in the directed graph where the processing function f works abnormally according to the input signal x, output signal y, and the value of the monitoring parameter S; S4, traversal and positioning: After completing the conversion of the directed graph in S3, starting from the input node, for each node in the directed graph, calculate its output signal y and the theoretical value of the observation parameter S according to the value of all the input signals x of the node, and compare the calculated theoretical value with the actual value obtained by actual observation. If there is a difference, it indicates that a fault has occurred at this node, and fault diagnosis and positioning are achieved. Otherwise, continue to calculate the next node until the traversal calculation of all nodes is completed.

2. A model-based measurement and control equipment fault diagnosis method according to claim 1, characterized in that: When searching for abnormal nodes in S4, the processing function f of each node in the measurement and control equipment system is given. i , observation function G i , and given the external input signal x of the measurement and control equipment system j , the measurement and control equipment system outputs signals to the outside j , and monitoring parameters of the measurement and control equipment system k .

3. A model-based measurement and control equipment fault diagnosis method according to claim 1, characterized in that: In S4, the theoretical value of the output signal y is calculated by processing the function f.

4. A model-based measurement and control equipment fault diagnosis method according to claim 1, characterized in that: In S4, the theoretical value of the monitoring parameter S is calculated by the observation function G.

5. A model-based measurement and control equipment fault diagnosis method according to claim 2, characterized in that: Since the intermediate results are not available, the location of the faulty node cannot be determined by the processing function f alone, and the observation function G is introduced for joint judgment.

6. A model-based measurement and control equipment fault diagnosis method according to claim 1, characterized in that: The state monitoring parameters of any node are divided into two categories. One category is the parameters that directly describe the state of the node itself, which are represented by the observation function G; the other category is the parameters that describe the input signal of the node, which correspond to the output variables in the signal transmission process of any node. If the state is normal, it means that all links of the signal from the input to the node are normal; otherwise, it indicates that there is a fault point in the signal transmission process.

7. A model-based measurement and control equipment fault diagnosis method according to claim 1, characterized in that: When looking for nodes in the directed graph where the processing function f works abnormally, the credibility analysis is performed on the input signal x, the output signal y, and the values ​​of the monitoring parameter S, and the contribution rate D of the x, y, and S values ​​at a certain node is calculated. The larger the contribution rate, the higher the probability that the node causes equipment failure. The calculation steps of the contribution rate D are: S31: Collect node samples of input signal x, output signal y and monitoring parameter S; S32: Calculate the mean of the above node samples and S33: Calculate the characteristic value λx of each node sample i ,λy i and λS i ; S34: Calculate the contribution rate D of the x, y, and S values ​​at a certain node.

8. A model-based measurement and control equipment fault diagnosis method according to claim 7, characterized in that: The value of the input signal x in all nodes can be expressed as x=[x1, x2, x3, ..., x n ], the value of the output signal y in all node samples can be expressed as y = [y1, y2, y3, ..., y n ], the value of the monitoring parameter S in all node samples can be expressed as S = [S1, S2, S3, ..., S n ]; then the mean of the input signal x is The mean of the output signal y is The mean value of the monitoring parameter S is Eigenvalue Eigenvalue Eigenvalue The contribution rates D of x, y, and S values ​​at a certain node are respectively Where i=1,2,3,…,n.

9. A model-based measurement and control equipment fault diagnosis method according to claim 8, characterized in that: When D i The larger the value, the higher the probability that the node causes equipment failure and the higher the credibility. When troubleshooting, equipment problems at this node can be prioritized.

Images