Time-given discrete event system fault diagnosis method based on tag time Petri network

Abstract

The invention belongs to the technical field of industrial manufacturing, and particularly relates to a tag time Petri net-based time-assigned discrete event system fault diagnosis method, which comprises the following steps of: S1, modeling a time-assigned discrete event system by using a tag time Petri net; s2, on the basis of a time-label sequence (TLS), a modified state class graph (MSCG) is constructed, and a transition sequence which is logically consistent with the TLS is found out; s3, proposing a method based on linear programming by utilizing time constraints related to a transition sequence consistent with the TLS logic in the observable MSCG so as to judge the transition sequence consistent with the TLS time; and S4, according to the transition sequence consistent with the TLS time, judging whether the system has a fault or not. The invention provides a tag time Petri network system fault diagnosis method based on a tag time Petri network, which solves the problems in the background technology, reduces the calculation complexity and is more suitable for a complex system.

Description

Technical Field

[0001] This invention belongs to the field of industrial manufacturing technology, specifically relating to a fault diagnosis method for time-specified discrete event systems based on labeled time Petri nets. Background Technology

[0002] State estimation for timed discrete event systems (DES) is crucial for fault diagnosis and control of real-time systems such as intelligent manufacturing systems. Currently, the main modeling tools for timed DES are automata and Petri nets (PNs). Brandin and Wonham proposed a timed automata model by introducing tick events into the automata; the size of this model increases with the number of tick events. On the other hand, researchers have characterized the behavior of timed DES by adding time constraints to the transitions of PNs, proposing timed PNs and time PNs (TPNs). Because the transmit delay of TPN transitions has upper and lower bound constraints, and the time information is represented by real variables, the system model is more compact and concise; therefore, TPNs are widely used in real-time system modeling and control.

[0003] In state estimation of TPN systems, state space compression and computation are crucial. Existing state class graphs (SCGs) and partial order techniques can analyze the state space of TPN systems relatively well, but neither preserves the allowed emission intervals of transitions. In existing techniques, a modified state class graph (MSCG) is proposed to describe the allowed emission intervals of transitions by adding symbolic variables to the edges of the SCG, and this graph is used for state estimation of partially observable labeled TPN systems.

[0004] Research on fault diagnosis in TPN systems has yielded several advancements. Existing techniques construct the Fault Diagnosis Graph (FDG) by removing nodes associated with unobservable transitions from the SCG, and propose an online fault diagnosis algorithm based on the FDG. This algorithm reduces computational complexity by not requiring the computation of the entire system state space and updating the FDG associated with observable transitions only after each observation. Another approach is the partial observable TPN fault diagnosis method based on the MSCG, which performs online checks on offline-computed MSCGs, providing a sound theoretical basis for the diagnosability of practical systems. Furthermore, existing techniques propose the concepts of K-diagnosability and τ-diagnosability for labeled TPN systems, including determining whether individual fault events can always be deterministically identified after a finite time. These techniques extend such analyses to more complex behaviors and propose an efficient method for checking diagnosability using model testing techniques. However, the problem of efficient fault diagnosis in time-based DES based on TPN remains unresolved.

[0005] Therefore, this invention proposes a fault diagnosis method for time-specified discrete event systems based on labeled time Petri nets. Summary of the Invention

[0006] The purpose of this invention is to provide a fault diagnosis method for time-assigned discrete event systems based on tagged time Petri nets (TTNs). This method reduces the cost of state estimation by constructing an observable MSCG associated with the TLS (Transition Time Limits). Secondly, based on solving the LPP (Low-Progression Point) associated with the transition timing constraints in the observable MSCG of the tagged TPN system, a method for calculating a time-consistent LTS (Long-Term Time Limits) set with TLS time is proposed. Finally, using these consistent LTS, the occurrence of a fault can be accurately determined, reducing computational complexity and making it more applicable to complex systems.

[0007] The specific technical solution adopted by this invention is as follows:

[0008] A fault diagnosis method for time-specified discrete event systems based on labeled time Petri nets includes the following steps:

[0009] S1: Modeling a timed discrete event system using a labeled time Petri net;

[0010] S2. Construct an observable modified state class graph (MSCG) based on the time-label sequence (TLS) and find the transition sequence that is logically consistent with the time-label sequence (TLS).

[0011] S3. By utilizing the time constraints related to transition sequences that are logically consistent with the time-labeled sequence TLS in the observable modified state class graph MSCG, a linear programming-based method is proposed to determine the transition sequences that are time-consistent with the time-labeled sequence TLS.

[0012] S4. Determine whether the system has failed based on the transition sequence that is consistent with the TLS time stamp sequence.

[0013] The technical effects achieved by this invention are as follows:

[0014] (1) This invention proposes a fault diagnosis scheme for time-discrete event systems, which greatly reduces the production cost and failure rate of enterprises and is conducive to maximizing enterprise benefits.

[0015] (2) The present invention proposes an observable modified state class graph (MSCG) construction algorithm for representing part of the state space of a tag-time Petri net system, which reduces the size of the original modified state class graph (MSCG) and lowers the computational cost of state estimation.

[0016] (3) Based on the observable modified state class diagram MSCG, this invention can quickly search for all transition sequences that are consistent with the time tag sequence TLS logic.

[0017] (4) In this invention, by analyzing the satisfaction of the time constraint set related to the transition in the observation modified state class graph MSCG, an estimation method for time consistent transition sequence is proposed to determine whether the system has failed, thereby reducing the computational complexity and making it more suitable for complex systems. Attached Figure Description

[0018] Figure 1 This is a flowchart illustrating an embodiment of a time-based discrete event system fault diagnosis method based on a labeled time Petri net according to the present invention.

[0019] Figure 2 This is a schematic diagram of a flexible manufacturing system part processing unit according to an embodiment of the fault diagnosis method for time-specified discrete event systems based on labeled time Petri nets of the present invention.

[0020] Figure 3 This is a schematic diagram of a tag-time Petri net model, representing an embodiment of a time-based discrete event system fault diagnosis method based on tag-time Petri nets according to the present invention.

[0021] Figure 4 This is a schematic diagram of a Tag-Time Petri Net Observable MSCG, representing an embodiment of a time-based discrete event system fault diagnosis method based on Tag-Time Petri Nets according to the present invention. Detailed Implementation

[0022] To make the objectives and advantages of this invention clearer, the invention will be specifically described below with reference to embodiments. It should be understood that the following text is merely used to describe one or more specific embodiments of the invention and does not strictly limit the scope of protection specifically claimed by the invention.

[0023] like Figure 1 As shown, a fault diagnosis method for a time-based discrete event system based on labeled time Petri nets includes the following steps:

[0024] S1: Modeling a timed discrete event system using a labeled time Petri net;

[0025] S2: Construct an observable modified state class graph (MSCG) based on the time-labeled sequence (TLS) to find the transition sequence that is consistent with the TLS logic;

[0026] S3: By utilizing the time constraints associated with the logically consistent transition sequences in the observable MSCG, a linear programming-based method is proposed to determine the time-consistent transition sequences with TLS.

[0027] S4: Determine whether the system has failed based on the transition sequence consistent with the TLS time.

[0028] In S1, the label-time Petri net consists of triples N t = (N,Q,L) represents a Petri net, where N = (P,T,Pre,Post) represents the location set P, which represents the space where resources are stored in the manufacturing system; the transition set T represents each process in the manufacturing system; Pre (Post): These are the pre- or post-association functions of N, respectively; It is a change t i Assign two nonnegative rational numbers l i and u i The time function, and Q(t) i )=[l i ,u i ] represents t i The static emission interval, i.e., the processing time of the process; L:T→{ε}∪Λ is the label function associated with the transition, which specifies an empty string ε or a symbol belonging to the label Λ set for each transition t∈T. u ={t∈T|L(t)=ε} denotes the set of unobservable transitions, T o ={t∈T|L(t)∈Λ} denotes the observable transition set. The notation for a Petri net is M = [M(p1)M(p2)…M(p...].m )] T M represents the resource configuration of all repositories.

[0029] In S2, the time transition sequence of the tag-time Petri net system is represented as follows: Its meaning is the sequence of processes and the corresponding completion time of each process. The logical transition sequence of σ is represented as follows: The observable projection is denoted as The set of transitions enabled at M is denoted as E(M) = {t∈T|M≥Pre(·,t)}. Enabling transition t at M generates a new identifier M′ = M + C(·,t), where C(·,t) is the column vector in the Petri net's incidence matrix C = post-Pre associated with t∈T. A time transition sequence σ is emitted from M0 to generate a reachable identifier M. k The process is denoted as M0[σ>M k A time-stamped sequence (TLS) is represented as ω = (α1,β1)(α2,β2)…(α…). h ,β h ), The logical label sequence represents ω. The state representation of the label-time Petri net system is Z. k =(M k ,Φ k ), where M k It is an reachability indicator. The inequalities in the equation show the transition t i ∈E(M k After Launched after a time unit, and after a period of time Launched a time unit ago. If the transition t... i Satisfying M≥Pre(·,t) i and time constraints in It is a change j ∈E(M k In M k The upper bound of t is then i Launchable. In Z k The set of transitions that can be emitted is denoted as

[0030] The observable modified state class graph (MSCG) of a labeled time Petri net system is a directed graph in which nodes represent state classes, denoted as S. k =(M k ,Θ k ), where M k For reachability indicators, It is a set of changes t i ∈E(M k Inequalities related to emission delay; edges connecting state classes are marked as in Indicates from S j The target class t to the edge i The allowed transmission interval. The observable MSCG algorithm specifically includes the following steps:

[0031] Input: Label-time Petri net, S cr =(M cr ,Θ cr ) and TLSω=(α1,β1)(α2,β2)…(α h ,β h );

[0032] S21: Initialize the set ω * :={α1,α2,…,α h};

[0033] S22: If the condition is met and From ω * Select an observation label α j ;

[0034] S23: If the conditions are met Then from Select a state class S from the list. k ;

[0035] S24: For all satisfying L(t) i )=α j or t i ∈T u Changes Calculate M q =M k +C(·,t i );

[0036] S241: For all enabling transitions t d ∈E(M q If condition t is satisfied d ∈E(M k ) and M k -Pre(·,t i )≥Pre(·,t d ),but otherwise

[0037] S242: Calculate S q =(Mq ,Θ q ),in And add S q To Ψ t ;

[0038] S243: If Ψ t There is no S q State class S other than e Equivalent to S q Then from Ψ t Move out of S q And add a line from S k To S e And marked as t i ,L(t i ), The edge;

[0039] S244: If Ψ t There is no S in the middle. q State class S other than e Equivalent to S q Then add a line from S k To S q And marked as t i ,L(t i ), The edge;

[0040] S2441: If t i ∈T u ,but

[0041] S2442: If but

[0042] S25: From Move out of S k Then proceed to step S23 to consider the next state class;

[0043] S26: From ω * Remove α from the middle j ;make and like and Then proceed to step S23; otherwise, proceed to step S22 to consider the next observation label.

[0044] Output: Observable MSCG.

[0045] In S2, a labeled time Petri net system and a time label sequence ω = (α1,β1)(α2,β2)...(α...) are given. h ,β hThe set of transition sequences consistent with ω logic is:

[0046]

[0047] In S3, a Petri net system with given labels and a time-labeled sequence ω = (α1,β1)(α2,β2)...(α) is observed at time τ. h ,β h The set of transition sequences consistent with time ω is defined as follows:

[0048]

[0049] in express The launch time of the last transition in the process. express In M k The actual transmission delay at the location, if the actual transmission delay is no greater than Changes Then the change occurs at time... The pre-emission causes the system to generate a new identifier. The algorithm for estimating the transition sequence set consistent with time ω includes the following steps:

[0050] Input: Tag-time Petri net system, S cr =(M cr ,Θ cr ) and the TLSω observed at time τ = (α1,β1)(α2,β2)…(α h ,β h );

[0051] S31: Initialization

[0052] S32: Construct an observable MSCG and find T(ω); Let ∏(ω) be all elements in the observable MSCG that satisfy... path A set;

[0053] S33: For each path Construct a constraint set that is related to the observation time;

[0054] S331: If the corresponding linear programming problem (LPP)(5) has a feasible solution, then

[0055] S332: If in the final state class There are unobservable changes (i.e.) If LPP(6) has a feasible solution, then

[0056] Output: Transition sequence set T(ω,τ).

[0057] In step S33, after obtaining the transition sequence T(ω) consistent with the logic of ω, the satisfiability of the time constraints related to the transitions is analyzed by solving the LPP. The set of transition sequences T(ω,τ) consistent with the ω-time observed at time τ is calculated, and a set of transition sequences consistent with the observable MSCG is defined. The relevant paths are as follows:

[0058]

[0059] in, And l = 1, 2, ..., q.

[0060] logical transition sequence Time consistency extended to path The estimation method for time-consistent transition sequences includes the following three steps:

[0061] Step A: Extraction and Path And the time-stamped sequence ω=(α1,β1)(α2,β2)…(α) observed at time τ h ,β h The associated constraints are shown in equation (4):

[0062]

[0063] Step B: Construct a linear programming problem. If the following LPP has a feasible solution:

[0064] Then the path Consistent with ω logic.

[0065] Step C: Check the given time τ and arrival Is the difference in time consumed less than in The minimum value of the upper bound of the triggerable transition (i.e. Consider a path that is logically consistent with the observed TLSω. If the following LPP(6) does not have a feasible solution:

[0066]

[0067] If constraint (6b2) is violated, then the path... In Consistent with ω time. If constraint (6b2) is satisfied, then a feasible solution of LPP(6) will lead to the system triggering an unobservable transition. From the last state class Reach its subsequent state class.

[0068] S4 includes: determining whether a system failure has occurred based on a transition sequence consistent with TLS time. The determination algorithm includes the following steps:

[0069] Input: Tag-time Petri net system, S cr =(M cr ,Θ cr ) and fault transition set T F ;

[0070] S41: Initialization

[0071] S42: If at time τ, TLSω=(α1,β1)(α2,β2)…(α h ,β h If the condition is met, proceed with the next steps; otherwise, proceed with all...

[0072] S43: Calculate T(ω,τ);

[0073] S44: For all Perform the following judgment;

[0074] S441: If but

[0075] S442: If there exists a Make And there exists a Make but

[0076] S443: If but

[0077] Output: Fault diagnosis status

[0078] In S4, given a labeled time Petri net system, a time label sequence ω = (α1,β1)(α2,β2)…(α…). h ,β h ), and a fault transition set T F The diagnostic tool is a function. It will include ω, τ and Associated with a diagnostic status, where It is TLSω=(α1,β1)(α2,β2)…(α h ,β hLet Nm, Un, and Fa be a set of states, where Nm, Un, and Fa represent normal, uncertain, and faulty states, respectively. Given a fault transition... If for all satisfy but If there exists a Make And there exists a Make but If for all but in, The transition set in the middle is represented as

[0079] This invention provides a fault diagnosis method for time-assigned discrete event systems based on labeled time Petri nets. This method offers a fault diagnosis scheme for real-time systems, reducing production costs and improving the economic efficiency of real-time systems. The code implementations of the observable MSCG construction algorithm, LTS estimation algorithm, and fault diagnosis algorithm can be generated using Python and run in the PyCharm interpreted environment, or implemented using other software tools such as Matlab; no limitation is made here. Furthermore, the observable MSCG construction algorithm, LTS estimation algorithm, and fault diagnosis algorithm used in this invention are not limited to the embodiments provided in this method, but can also be applied to other embodiments.

[0080] In actual operation, this invention, such as Figure 1 As shown, in this embodiment of the invention, the fault diagnosis method for a time-based discrete event system based on a labeled time Petri net includes the following steps:

[0081] S1. Modeling a timed discrete event system using a labeled time Petri net;

[0082] S2. Construct an observable-modified state class graph (MSCG) based on the time-labeled sequence (TLS) and find the transition sequence that is consistent with the TLS logic;

[0083] S3. By utilizing the time constraints associated with the transition sequences that are logically consistent with TLS in the observable MSCG, a linear programming-based method is proposed to determine the transition sequences that are time-consistent with TLS.

[0084] S4. Determine whether the system has failed based on the transition sequence consistent with the TLS time.

[0085] The following is combined Figure 2 and Figure 3 Step S1 in this implementation will be explained.

[0086] In Example 1, a flexible manufacturing system parts processing unit comprises a computer numerical control (CNC) machine tool and a material handling system. It includes two machine tools M1 and M2, two overhead racks H1 and H2, and a rotary robot R. There is no limit to the number of workpieces that can be processed on each machine tool, nor is there a limit to the number of workpieces that the robot can hold at one time. Figure 2 As shown, this system produces two types of workpieces, W1 and W2. Workpieces enter and exit the system through two loading / unloading ports, I1 / O1 and I2 / O2, with a robot responsible for their movement. Ultimately, the system's process flow is as follows: Figure 3 The label-time Petri net model is used in the model, where the physical meanings of the collection set P = {p1,p2,p3,p4,p5,p6,p7} and the transition set T = {t1,t2,t3,t4,t5,t6} are given in Table 1.

[0087] Table 1. Physical meaning of locations and changes

[0088]

[0089] The pre- and post-correlation functions of the label-time Petri net are as follows:

[0090]

[0091] The time functions of the transitions are Q(t1) = [0,4], Q(t2) = [1,3], Q(t3) = [0,4], Q(t4) = [0,3], Q(t5) = [1,5], Q(t6) = [0,3]; the observable transition set is T. o = {t1, t4, t6}, with labels L(t1) = a, L(t4) = b, and L(t6) = c respectively; the unobservable transition set is T. o ={t2,t3,t5}, ​​with labels L(t2) = L(t3) = L(t5) = ε respectively. Fault transition set T F ={t3}. The initial identifier of the system is M0 = [1001001] T .

[0092] The following is combined Figure 3 , Figure 4 Step S2 in this implementation will be explained.

[0093] If the tag-time Petri net system is in the initial state Z0=(M0,Φ0)(M0=[1001001]), T If Φ0 = {0 ≤ φ1 ≤ 4, 0 ≤ φ4 ≤ 3}, then the current state class S crEqual to S0, the emittable transitions are t1 and t4. Assuming that TLSω = (a,1)(b,5)(c,6) is observed at time τ = 9, then during the initialization of the observable MSCG algorithm, ω * ={a,b,c}. Where, This is used to store the state classes generated by emitting observable transitions labeled a, b, and c. To reduce computational cost, this invention only calculates the new identifier M generated by emitting specified transitions (i.e., non-observable transitions or observable transitions labeled a, b, and c). q =M k +C(·,t i ).

[0094] In the initial state class, label a is selected first. Therefore, at S0, only the emission of transition t1 is considered, resulting in a new label M1 = M0 + C(·, t1), where C = Post-Pre. Next, S1 = (M1, Θ1) is calculated, where Φ1 = {1 ≤ φ2 ≤ 3}, and S1 is added to Ψ. t At this time, Ψ t Since there is no state class other than S1 in the {S0, S1} that is equivalent to S1, we can add a path from S0 to S1 and mark it as... The edge, of which Since S1 is obtained from the observable transition t1 of the emission, therefore At the same time, move S0 out but At this point, it is necessary to start from ω * By removing 'a' from the middle, we can obtain ω. * ={b,c}.

[0095] Furthermore, under state class S1, from ω * = {b, c}, label b is selected. Similarly, considering only the emission unobservable transitions and the observable transitions with label b, the new labels M2 = M1 + C(·, t2), M3 = M2 + C(·, t3), M4 = M2 + C(·, t4), and M5 = M3 + C(·, t4) are obtained, thus yielding state classes S2, S3, S4, and S5 respectively. Similarly, when label c is selected, considering only the emission unobservable transitions and the observable transitions with label c, state classes S6, S7, S8, S9, and S1 are calculated. 10 S 11 S 12 and S 13 Finally, by adding edges to the calculated state class, we can obtain the state with respect to ω. * The observable MSCG associated with the labels in {a,b,c}, such as Figure 4 As shown, the blue-marked edges represent the edges related to ω.* The observable transitions related to the observed labels are obtained. Furthermore, the set of transition sequences logically consistent with ω=(a,1)(b,5)(c,6) can be obtained as T(ω)={t1t2t3t4t5t6,t1t2t4t3t5t6,t1t2t4t5t6,t1t2t4t5t6,t1t2t4t5t6t3,t1t2t4t5t3t6}.

[0096] The following is combined Figure 3 , Figure 4 Step S3 in this implementation will be explained.

[0097] according to Figure 3 According to the observable MSCG, the transition sequence is logically consistent with ω=(a,1)(b,5)(c,6). The relevant path is

[0098]

[0099] By obtaining the time constraints associated with transition ω, as shown below:

[0100]

[0101] By introducing an objective function To constraint (7), and obtain a feasible solution, it means that the path π(t1t2t4t5t6) is logically consistent with TLSω=(a,1)(b,5)(c,6) observed at time τ=9.

[0102] because The path π(t1t2t4t5t6) should be determined by solving the following LPP. Does it coincide with the TLSω = (a,1)(b,5)(c,6) observed at time τ = 9?

[0103]

[0104] Since a solution can be found, This is inconsistent with the TLSω = (a,1)(b,5)(c,6) observed at time τ = 9. Similarly, all LTSs consistent with the ω observed at time τ = 9 can be found, i.e., T(ω,τ) = {t1t2t3t4t5t6,t1t2t4t3t5t6,t1t2t4t5t6,t1t2t4t5t6t3,t1t2t4t5t3t6}. Finally, it can be known that D(ω,τ,t3) = Fa.

[0105] Therefore, this invention provides a fault diagnosis method for time-consistent discrete event systems based on labeled time Petri nets, which solves the problems in the background technology. By analyzing the satisfaction of the time constraint set related to transitions in the observable MSCG, it proposes an estimation method for time-consistent transition sequences to seek the transition sequence with the minimum cost, thereby reducing the computational complexity and making it more suitable for complex systems.

[0106] The above description is merely a preferred embodiment of the present invention. It should be noted that those skilled in the art can make various improvements and modifications without departing from the principles of the present invention, and these improvements and modifications should also be considered within the scope of protection of the present invention. Structures, devices, and operating methods not specifically described or explained in this invention are implemented according to conventional methods in the art unless otherwise specified or limited.

Claims

1. A fault diagnosis method for a time-specified discrete event system based on labeled time Petri nets, characterized in that, Includes the following steps: S1: Modeling a timed discrete event system using a labeled time Petri net; S2: Construct an observable modified state class graph based on the time-labeled sequence, and find the transition sequence that is logically consistent with the time-labeled sequence; S3: By utilizing the time constraints related to transition sequences that are logically consistent with the time-labeled sequence in the observable modified state class diagram, a linear programming-based method is proposed to determine the transition sequences that are time-consistent with the time-labeled sequence. S4: Determine whether the system has failed based on the transition sequence that is consistent with the time of the time label sequence.

2. The fault diagnosis method for a time-based discrete event system based on a labeled time Petri net, as described in claim 1, is characterized in that: In S1, the label-time Petri net consists of triples N t = (N,Q,L) represents the Petri net set, where N = (P,T,Pre,Post) represents the set of resources in the manufacturing system, P represents the space where resources are stored, and T represents the various processes in the manufacturing system. These are the pre- or post-association functions of N, respectively; It is a change t i Assign two nonnegative rational numbers l i and u i The time function, and Q(t) i )=[l i ,u i ] represents t i The static emission interval is the processing time of the process; L:T→{ε}∪Λ is the label function associated with the transition. The label function specifies an empty string ε or a symbol belonging to the label Λ set for each transition t∈T. T u ={t∈T|L(t)=ε} denotes the set of unobservable transitions, T o ={t∈T|L(t)∈Λ} represents the observable transition set, and the Petri net is denoted as M=[M(p1)M(p2)…M(p... m )] T M represents the resource configuration of all repositories.

3. The fault diagnosis method for a time-based discrete event system based on a labeled time Petri net, as described in claim 2, is characterized in that: In S2, the time transition sequence of the tag-time Petri net system is represented as follows: Its meaning is the sequence of processes and the corresponding completion time of each process; the logical transition sequence of σ is represented as follows: The observable projection is denoted as The set of transitions enabled at point M is represented as E(M) = {t ∈ t | M ≥ Pre(·, t)}; the emission of the enabling transition t at point M will produce a new identifier M. ′ =M+C(·,t), where C(·,t) is the column vector associated with t∈T in the Petri net's incidence matrix C=Post-Pre; a time-varying sequence σ is emitted from M0 to generate the reachability identifier M. k The process is denoted as M0[σ>M k A time-stamped sequence (TLS) is represented as ω = (α1, β1)(α2, ​​β2)...(α...). h ,β h ), The logical label sequence representing ω; the state representation of the label-time Petri net system is Z. k =(M k ,Φ k ), where M k It is an reachability indicator. The inequalities in the equation show the transition t i ∈E(M k After Launched after a time unit, and after a period of time Launched a time unit ago; if the transition t i Satisfying M≥Pre(·,t) i and time constraints in It is a change j ∈E(M k In M k The upper bound of t is then i Launchable; in Z k The set of transitions that can be emitted is denoted as 4. The fault diagnosis method for a time-based discrete event system based on a labeled time Petri net as described in claim 3, characterized in that: In S2, the observable modified state class graph of the tag-time Petri net system is a directed graph, where each node represents a state class, denoted as S. k =(M k ,Θ k ), where M k For reachability indicators, It is a set of changes t i ∈E(M k Inequalities related to emission delay; edges connecting state classes are marked as "t". i ,L(t i ), ",in Indicates from S j The target class t to the edge i The allowed emission interval; the observable correction state class graph algorithm specifically includes the following steps: Input: Label-time Petri net, S cr =(M cr ,Θ cr ) and TLSω=(α1,β1)(α2,β2)…(α h ,β h ); S21: Initialize the set Ψ t :={S cr }; ω * :={α1,α2,…,α h }; S22: If the condition is met and From ω * Select an observation label α j ; S23: If the conditions are met Then from Select a state class S from the list. k ; S24: For all satisfying L(t) i )=α j or t i ∈T u Changes Calculate M q =M k +C(·,t i ); S241: For all enabling transitions t d ∈E(M q If condition t is satisfied d ∈E(M k ) and M k -Pre(·,t i )≥Pre(·,t d ),but otherwise S242: Calculate S q =(M q ,Θ q ),in And add S q To Ψ t ; S243: If Ψ t There is no S q State class S other than e Equivalent to S q Then from Ψ t Move out of S q And add a line from S k To S e And marked as t i ,L(t i ), The edge; S244: If Ψ t There is no S in the middle. q State class S other than e Equivalent to S q Then add a line from S k To S q And marked as t i ,L(t i ), The edge; S2441: If t i ∈T u ,but S2442: If but S25: From Move out of S k Then proceed to step S23 to consider the next state class; S26: From ω * Remove α from the middle j ;make and like and Then proceed to step S23; otherwise, proceed to step S22 to consider the next observation label. Output: Observable modified state class graph MSCG.

5. The fault diagnosis method for a time-based discrete event system based on a labeled time Petri net, as described in claim 4, is characterized in that: In S2, a labeled time Petri net system and a time label sequence ω = (α1,β1)(α2,β2)...(α...) are given. h ,β h The set of transition sequences consistent with ω logic is:

6. The fault diagnosis method for a time-based discrete event system based on a labeled time Petri net, as described in claim 5, is characterized in that: In S3, a Petri net system with given labels and a time-labeled sequence ω = (α1,β1)(α2,β2)...(α) is observed at time τ. h ,β h The set of transition sequences consistent with time ω is defined as follows: in express The launch time of the last transition in the process. express In M k The actual transmission delay at the location, if the actual transmission delay is no greater than Changes Then the change occurs at time... The pre-launch process causes the system to generate a new identifier; The algorithm for estimating a set of transition sequences consistent with time ω includes the following steps: Input: Tag-time Petri net system, S cr =(M cr ,Θ cr ) and the TLSω observed at time τ = (α1,β1)(α2,β2)…(α h ,β h ); S31: Initialization S32: Construct an observable MSCG and find T(ω); Let Π(ω) be all elements in the observable MSCG that satisfy... path A set; S33: For each path Construct a constraint set that is related to the observation time; S331: If the corresponding linear programming problem (LPP)(5) has a feasible solution, then S332: If in the final state class There are unobservable changes (i.e.) If LPP(6) has a feasible solution, then Output: Transition sequence set T(ω,τ).

7. The fault diagnosis method for a time-based discrete event system based on a labeled time Petri net, as described in claim 6, is characterized in that: In step S33, after obtaining the transition sequence T(ω) consistent with the ω logic, the satisfiability of the time constraints related to the transitions is analyzed by solving the LPP, and the set of transition sequences T(ω,τ) consistent with the ω time observed at time τ is calculated; based on the observable modified state class graph MSCG, a [missing information - likely a class name or structure] is defined. The relevant paths are as follows: in, and l = 1, 2, ..., q; logical transition sequence Time consistency extended to path The estimation method for time-consistent transition sequences includes the following three steps: Step A: Extraction and Path And the time-stamped sequence ω=(α1,β1)(α2,β2)…(α) observed at time τ h ,β h The associated constraints are shown in equation (4): Step B: Construct a linear programming problem. If the following LPP has a feasible solution: Then the path Consistent with ω logic; Step C: Check the given time τ and arrival Is the difference in time consumed less than in The minimum value of the upper bound that can trigger a transition; that is Consider a path that is consistent with the observed TLSω logic. If the following LPP(6) does not have a feasible solution: If constraint (6b2) is violated, then the path... In Consistent with ω time; if constraint (6b2) is satisfied, then a feasible solution of LPP(6) will lead to the system triggering an unobservable transition. From the last state class Reach its subsequent state class.

8. The fault diagnosis method for a time-based discrete event system based on a labeled time Petri net, as described in claim 7, is characterized in that: S4 includes: determining whether a system failure has occurred based on a transition sequence consistent with the TLS time stamp sequence; the determination algorithm includes the following steps: Input: Tag-time Petri net system, S cr =(M cr ,Θ cr ) and fault transition set T F ; S41: Initialization S42: If at time τ, TLSω=(α1,β1)(α2,β2)…(α h ,β h If the condition is met, proceed with the next steps; otherwise, proceed with all... S43: Calculate T(ω,τ); S44: For all Perform the following judgment; S441: If but S442: If there exists a Make And there exists a Make but S443: If but Output: Fault diagnosis status 9. The fault diagnosis method for a time-based discrete event system based on a labeled time Petri net as described in claim 8, characterized in that: In S4, given a labeled time Petri net system, a time label sequence ω = (α1,β1)(α2,β2)…(α…). h ,β h ), and a fault transition set T F The diagnostic tool is a function. It will include ω, τ and Associated with a diagnostic status, where It is TLSω=(α1,β1)(α2,β2)…(α h ,β h A set of Nm, Un, and Fa, where Nm, Un, and Fa represent normal, uncertain, and fault states, respectively; given a fault transition. If for all satisfy but If there exists a Make And there exists a Make but If for all but in, The transition set in the middle is represented as

Images