Data loss error detection method and system in information physical system based on petri-net

Abstract

To provide a data loss error detection method and system in an information physical system based on a Petri-Net.SOLUTION: A method comprises the steps of: firstly, mapping an information physical system into a Petri-Net with data for generating a merging process; optimizing events and conditions in the merging process based on a deletion rule, and obtaining an optimized merging process; determining whether or not each event is a loss event in the optimized merging process; and outputting an event set of data loss errors. By optimizing the merging process based on the deletion rule, complexity of a model is reduced, a data loss error detection algorithm is designed, and the data loss error in information physical system data stream is detected, allowing verification of design correctness of the information physical system.SELECTED DRAWING: Figure 2

Description

[Technical Field]

[0001] The present invention relates to the Petri-Net data detection field, particularly to a method and system for detecting data loss errors in information physics systems. [Background technology]

[0002] Computational physics systems (CPSs) are multidimensional complex systems that integrate computing, networks, and the physical environment. They enable real-time sensing, dynamic control, and information processing for large-scale engineering systems through the deep integration and collaboration of computational intelligence, communication, and control. CPSs can improve industrial efficiency, quality, safety, and reliability, reduce costs and resource consumption, and create new value and business. Currently, CPSs are widely applied in industrial fields such as intelligent power grids, intelligent transportation, and intelligent manufacturing. However, with the increase in distributed sensors and parallel processing units, CPS processing processes are becoming increasingly complex. Therefore, how to ensure the accuracy of CPS design—that is, how to ensure that CPSs can execute according to their design goals and requirements without errors or anomalies—has become a hot research topic in both industry and academia.

[0003] Business Process Modeling Notation (BPMN) is a standardized notation for business processes and is widely used to model the control streams and data streams of CPSs. However, due to the lack of a formalization method, verifying the design accuracy of a CPS, especially its data streams, is difficult. Currently, research on detecting data errors in CPS data streams mainly includes test-based and model-based methods. Petri-Net-based methods are model-based methods that typically detect data stream errors using reachable graphs (RGs) or state transition graphs. For example, a system systematically traverses a given workstream to detect its data stream errors. However, traditional methods are prone to state space explosion, have complex analysis structures, and have low detection efficiency. Summary of the Invention [Problem to be solved by the invention]

[0004] Object of the invention: The object of the present invention is to provide a method and system for detecting data loss errors in information physics systems based on Petri-Net, which has low model complexity and high detection efficiency. [Means for solving the problem]

[0005] Technical solution: A data loss error detection method in an information physics system based on Petri-Net according to the present invention, comprising: Mapping an information-physics system modeled in the business process modeling notation to a first Petri net that is a Petri net with data; generating a merge process μ=(B, E, G, D, Rd, Wr, De) of the first Petri net, where B is a condition set, E is an event set, G is an arc set, D is a data element set, Rd is a read marking function, Wr is a write marking function, and De is a delete marking function; optimizing the events and conditions in the merging process μ based on a deletion rule to obtain an optimized merging process μ′=(B′,E′,G′,D′,Rd′,Wr′,De′,λ′,M′); In the optimized merging process μ′, determining whether each event is a lost event and outputting a set of events with data loss errors; a step of performing a synonymous data operation on the optimized merging process μ′ and further calculating data proximity events, the synonymous data operation being to provide a data write element d of an event e, [e] + If there are two parallel events e1 and e2 belonging to d∈Wr(e1)∪Rd(e1), then add the same data operation label, i.e., (e1,Wr:d or e1,Rd:d), from e1 to e2, where [e] +={e′∈E|e′≦e}, where the data proximity event is to provide two events e1 and e2 that satisfy the data d∈Rd(e1)∪Wr(e1)∪De(e1). When e1<e2, there is clearly a path from e1 to e2, and for an event e′ on this path

Number

[0006] Based on the data objects and states in the cyber-physical system, map the cyber-physical system to the first Petri net and include the step of modeling the first Petri net. The first Petri net is a data net N with an initial marker m0 where Σ=(N D ,m0), N D =(P,T,F,D,Read,Write,Delete), satisfying that (P,T,F) is a net, P is a library, T is a transition, F is a stream relationship, and Read, Write, and Delete are the read marking function, write marking function, and delete marking function in the first Petri net respectively. D

[0007] ​Furthermore, the step of mapping the information-physics system modeled in the business process modeling notation to the first Petri net, which is a Petri net with data, includes: The merge process μ=Merge(β) is included, and the method for generating the merge process μ is as follows: merge all β-conflict conditions with the same label and occurrence depth into one condition, Merge all β events with the same label, i.e., the before and after sets, into one event; All relevant arcs of β need to be redirected based on the merge conditions and events.

[0008] where β is the branching process of the first Petri net.

[0009] Furthermore, the step of mapping the information-physics system modeled in the business process modeling notation to the first Petri net, which is a Petri net with data, includes: mapping exclusive gateways of the business process modeling notation to a conflict structure of the first Petri net; mapping parallel gateways of the business process modeling notation to the parallel structure of the first Petri net; Mapping activities to transitions and their pre- or post-libraries; Mapping inputs and outputs of a data object or data library to marking functions of the data read / write elements during the transition; Mapping an external input to a marking function of the data writing element during the transition; Mapping message streams between activities to libraries and their pre- and post-transitions; Mapping other connections to arcs between the relevant libraries and transitions.

[0010] Furthermore, the step of optimizing the events and conditions in the merging process μ based on the deletion rule to obtain an optimized merging process μ′=(B′,E′,G′,D′,Rd′,Wr′,De′,λ′,M′) includes: The merging process μ is a target mutual merging process μ λ =(B,E,G,D,Rd,Wr,De,λ,M), where λ is a marking function used to mark the type of the current event, M:B→{0,1,2,3,...} is the identity of the net, and the marking function λ:E→L∪{α}, where L={!,?,@}×D includes non-marking α, write !, read ?, delete @; The target mutual merging process μ λ In this case, the rule deletion includes deleting events and conditions to obtain an optimized merging process μ′.

[0011] Furthermore, in the step of optimizing the events and conditions in the merging process μ based on the deletion rule to obtain an optimized merging process μ′=(B′,E′,G′,D′,Rd′,Wr′,De′,λ′,M′), The deletion rule is ∃e∈E∧λ(e)=α,

number

number

number

number

number

number

number

number

[0012] Furthermore, the step of optimizing the events and conditions in the merging process μ based on the deletion rule to obtain an optimized merging process μ′=(B′,E′,G′,D′,Rd′,Wr′,De′,λ′,M′) includes: The method includes a step of performing an accuracy analysis on the deletion rule, adopting weak reciprocity as a criterion for detecting whether behaviors are consistent, and reducing the problem of detecting whether the source merge process and the target reciprocal merge process maintain behavior consistency to the problem of detecting whether the source merge process and the target reciprocal merge process satisfy weak reciprocity.

[0013] Furthermore, in the optimized merging process μ′, the step of determining whether each event is a loss event and outputting a set of events of data loss errors may include: In the optimized merging process μ′, before determining whether each event is a lost event, the optimized merging process μ′ is initialized, and the initialization includes: Adding a condition b′0 and an event e′0 before the start condition of the optimized merging process μ′ and adding a condition b′ and an event e′ after the end condition, where the b′ and e′ indices are the maximum indices i of the conditions and events in μ′ plus 1, respectively; In the optimized merge process μ′, the stream relations {b′0}×{e′0}, {e′0}×{b0}, {b′ i}×{e′} and {e′}×{b′}; Marking e'0 as a delete operation for all data, i.e., De(e'0)=D, and marking e' as an end operation, i.e., End(e')=D, where End is an end marking function.

[0014] A data loss error detection system in an information physics system based on Petri-Net according to the present invention, comprising: a Petri net mapping unit for mapping an information-physics system modeled in the business process modeling notation to a first Petri net which is a data-attached Petri net; A merge process generation unit for generating the merge process μ = (B, E, G, D, Rd, Wr, De) of the first Petri net, where B is a set of conditions, E is a set of events, G is a set of arcs, D is a set of data elements, Rd is a read marking function, Wr is a write marking function, and De is a delete marking function. A merge process optimization unit for optimizing events and conditions in the merge process μ based on deletion rules to obtain an optimized merge process μ′ = (B′, E′, G′, D′, Rd′, Wr′, De′, λ′, M′). In the optimized merge process μ′, it includes a loss error detection unit for determining whether each event is a lost event and outputting a set of events of data loss error. In the loss error detection unit, Perform synonymous data operations on the optimized merge process μ′, and further calculate data proximity events. The synonymous data operation is to give the data write element d of event e, [e] + If there are two parallel events e1 and e2 belonging to [e] and d ∈ Wr(e1) ∪ Rd(e1), add the same data operation label, that is, (e1, Wr: d or e1, Rd: d) from e1 to e2, provided that [e] + ={e′ ∈ E|e′ ≦ e}, and the data proximity event is to give two events e1 and e2 that satisfy data d ∈ Rd(e1) ∪ Wr(e1) ∪ De(e1). When e1 < e2, there is clearly a path from e1 to e2, and for the event e′ on this path

Number

[0015] A computer-readable storage medium according to the present invention, on which a computer program is stored, is characterized in that, when the computer program is executed by a processor, the method for detecting data loss errors in an information physics system based on Petri-Net is realized. [Effects of the Invention]

[0016] Beneficial Effects: Compared with the prior art, the advantages of the present invention are as follows: (1) The present invention employs Petri net unfolding technology and, through structural analysis, alleviates the problem that traditional reachability graph or state transition graph detection methods suffer from state space explosion when dealing with large-scale systems, making it difficult to effectively detect data loss errors. (2) The present invention uses parallel relationship detection to compensate for the problem that, after a data stream error occurs, the state space of the reachability graph is incomplete, making it impossible to comprehensively detect data loss errors, thereby ensuring comprehensive detection. (3) The present invention designs an α-deletion rule and combines it with a merging process to reduce irrelevant control stream conditions and events while ensuring behavior consistency, thereby reducing model complexity and improving the efficiency of the data loss error detection algorithm. (4) In the present invention, a data loss error detection algorithm is designed to search for an event e belonging to the marking function of a data element d in its data proximity set to determine whether there is an event e′ belonging to the marking function of the data element d. Through the above steps, data loss errors in CPS data streams can be detected, and the design accuracy of the information physics system can be verified. Experimental comparisons also prove the advantages and effectiveness of the present invention. [Brief explanation of the drawings]

[0017] [Figure 1]FIG. 1 is a BPMN diagram of an embodiment of the present invention. [Figure 2] 2 is a flowchart of a data loss error detection method according to an embodiment of the present invention; [Figure 3] FIG. 1 is a diagram showing the mapping relationship between BPMN and PD-Net in an embodiment of the present invention. [Figure 4] 1 is a diagram and a reachability graph of α-deletion rule 1 according to an embodiment of the present invention. [Figure 5] 10 is a diagram and a reachability graph of α-deletion rule 2 according to an embodiment of the present invention. [Figure 6] 10 is a diagram and reachability graph of α-deletion rule 3 according to an embodiment of the present invention. DETAILED DESCRIPTION OF THE INVENTION

[0018] The technical solution will be further described below with reference to the drawings.

[0019] Referring to FIG. 1 and FIG. 2, in a CPS modeling process information physics system (CCPS) modeled using BPMN, the steps of performing data loss error detection on this CCPS using the present invention include the following steps.

[0020] S1,Based on the mapping relationship, CCPS modeled using BPMN is mapped to Petri-Net with data (PD-Net).

[0021] Considering the data objects and their states, we convert BPMN into PD-Net, and then use a modeling tool to model the PD-Net generated by the mapping on a computer. Here, PD-Net is a data-attached net (D-net) N with initial markers m0. D and Σ=(N D ,m0). D-net is a 7-tuple N D = (P, T, F, D, Read, Write, Delete), and (P, T, F) is a net. D is a finite set of data elements. Read: T → 2 Dis the marking function for data reading, Write:T→2 D is the marking function for writing data, Delete:T→2 D is the marking function for data deletion.

[0022] As shown in Figure 3, the mapping from BPMN to PD-Net is as follows: Map BPMN exclusive gateways to PD-Net collision structures. Map BPMN parallel gateways to PD-Net parallel structures. Map activities to transitions and their pre / post libraries. Map data object / data library inputs and outputs to marking functions of data read / write elements in transitions. Map external inputs to marking functions of data write elements in transitions. Map message streams between activities to libraries and their pre / post transitions. Map other connected objects to arcs between some related libraries and transitions.

[0023] S2,generates the PD-Net merging process according to the rules.

[0024] The formalization of the merge process (MP) μ = (B, E, G, D, Rd, Wr, De) maps the library P, transition T, stream relation F, data element set D, and marking functions Read, Write, and Delete to the condition B, event E, arc G, data element set D, and marking functions Rd, Wr, and De, respectively. The specific rules are as follows: Let β be the branching process of PD-Net(N, m0), and express it as μ = Merge(β), which can be obtained by the following steps.

[0025] (1) Merge all β-conflict conditions with the same label and occurrence depth into one condition.

[0026] (2) Merge all β events with the same label, i.e., the before and after sets, into one event.

[0027] (3) All related arcs of β need to be redirected based on merge conditions and events.

[0028] A branching process is a finite or infinite acyclic net with markers, which can be obtained by continuously triggering transitions and evolving a PD-Net under the following assumptions:

[0029] (1) From a set of libraries (called conditions), initial markers corresponding to each library are marked.

[0030] (2) Each new trigger generates a new transition (called an event).

[0031] (3) Generate a new library (also called conditions) for each newly generated token, where each event / condition is marked by a corresponding transition / condition.

[0032] If x is one of the nodes (condition or event) of β, then the occurrence depth of x is defined as the maximum number of marked nodes on any directed path in the directed graph of β that starts from the initial condition and ends at x.

[0033] Finally, we obtain the PD-Net merging process μ = (B, E, G, D, Rd, Wr, De).

[0034] S3, we create an α-elimination rule, optimize the generated merging process, and give an accuracy analysis of the α-elimination rule.

[0035] The α-elimination rule is to convert the merging process into a mutual MP and then optimize it to obtain the optimized merging process μ′, where μ λ=(B,E,G,D,Rd,Wr,De,λ,M), where the marking function λ:E→L∪{α} is, L={!,?,@}×D, marking the type of the current event including non-marking α, write !, read ?, and delete @, and M is the mapping M:P→{0,1,2,3,...}, called the identity of the net.

[0036] Alpha Removal Rule 1: ∃e∈E∧λ(e)=α,

number

number

[0037] The α-removal rule 1 can be applied to three cases as shown in Figure 4. Figures 4(a), 4(b), and 4(c) correspond to the first, second, and third cases, respectively, to which the α-removal rule 1 is applied. Figure 4(d) is the reachable graph corresponding to Figure 4(a), Figure 4(e) is the reachable graph corresponding to Figure 4(b), and Figure 4(f) is the reachable graph corresponding to Figure 4(c). To the left of the thick arrows, there is a source-reciprocal MP μ λ and this mutual formula MP μ λ There is a α in the source-optimized merge process (O-MP) μ′ to the right of the thick arrow. λIt is obtained by deleting α from . As shown in Figure 4(a), α deletion rule 1 is applied to a mutual MP with a serial structure. As shown in Figures 4(b) and 4(c), α deletion rule 1 is applied to a mutual MP with a parallel structure. The difference between Figures 4(a) and 4(c) is the different processing of M'. To visually distinguish α from data manipulation events, α is shown as a filled black block. Also, the area shown in the dashed block is the area where the source mutual MP μ is deleted during the α deletion process. λ It is a relevant element that needs to be removed from

[0038] Alpha removal rule 2: ∃e1,e2,...e n ∈E∧λ(e1)=α∧λ(e2)=α∧...∧λ(e n )=α,

number

number

number

[0039] The α-elimination rule 2 can be applied to two cases. As shown in Fig. 5, Fig. 5(a) and Fig. 5(b) correspond to the first and second cases to which the α-elimination rule 2 is applied, respectively. Fig. 5(c) is the reachable graph corresponding to Fig. 5(a), and Fig. 5(d) is the reachable graph corresponding to Fig. 5(b). The difference between Fig. 5(a) and Fig. 5(b) is the processing of M'. The same point is that the source reciprocal formula MP μ to the left of the thick arrow shown in Fig. 5(a) and Fig. 5(b) λ There are n αs in , and the inputs of these αs are gathered in condition b2, and the outputs are gathered in condition b3, and α-elimination rule 2 is used to eliminate all these αs.

[0040] Alpha Removal Rule 3: ∃e1,e2,...e n ∈E∧λ(e1)=α∧λ(e2)=α∧...∧λ(e n )=α∧λ(e)≠α,

number

number

[0041] The α-elimination rule 3 is shown in Figure 6. Figure 6(a) corresponds to the first case where the α-elimination rule 3 is applied, and Figure 6(b) is the reachable graph corresponding to Figure 6(a). The source reciprocal formula MP μ λ In the graph, there are n α and one data reading element z operation ?z, and the inputs of these events are gathered in condition b2, and the outputs are gathered in condition b3. If we directly delete all of these n α, the source mutual formula MP μ λTherefore, by deleting these n αs, the source mutual formula MP μ λ Simplifying the source reciprocal MP μ λ In order to maintain the selection (collision) structure in μ′, one α-event needs to be added to the objective-optimized merging process (O-MP) μ′.

[0042] Correctness analysis of the alpha deletion rule: We perform a correctness analysis on the α-deletion rule and adopt weak co-simulation as the criterion for detecting whether actions are consistent. The problem of whether the detected source MP and the target MP maintain action consistency can be reduced to the problem of detecting whether the detected source MP and the target MP satisfy weak co-simulation.

[0043] Based on the above α deletion rules, reduction optimization can be performed on the MP, which can reduce the number of control stream conditions and events, and greatly improve the efficiency of the data loss error detection algorithm under the premise of ensuring behavioral consistency.

[0044] The certification process includes the following steps:

[0045] S301, Accuracy Analysis of Alpha Removal Rule 1: The key issue in performing correctness analysis on α-deletion rule 1 is to prove whether the reachable graph of the source mutual formula MP and the reachable graph of the target mutual formula MP satisfy weak mutual simulation. The proof is as follows.

[0046] That is, we prove that the reachable graphs shown in Figure 4(d) satisfy mutual simulation. In the first step, the reachability graph on the left contains

number

[0047] In the second step, the reachability graph on the left has

number

[0048] As can be seen from the above, M0S0'. The relationship S = {(M0,M0'),(M1,M0'),(M2,M1')}.

[0049] Reverse S of S relationship -1 ={(M0′,M0),(M0′,M1),(M1′,M2)} is a weak simulation, i.e., the reachable graph on the right of the thick arrow in Figure 4 is weakly simulated by the reachable graph on the left.

[0050] In the first step, the right reachable graph has M0′ unchanged, and the left reachable graph has M0.

number

[0051] In the second step, the reachable graph on the right has

number

[0052] As can be seen from the above, M0′S -1 M0. From this, we prove that M0 ≒ M0′. Similarly, we can prove that other reachable graphs satisfy weak mutual simulation.

[0053] S302, Accuracy Analysis of Alpha Removal Rule 2: The key issue in analyzing the accuracy of α-deletion rule 2 is to prove whether the reachable graph of the source mutual formula MP and the reachable graph of the target mutual formula MP satisfy weak mutual simulation. The proof is as follows.

[0054] That is, we prove that the two reachable graphs shown in Figure 5 satisfy weak mutual simulation.

[0055] In the first step, the reachability graph on the left contains

number

[0056] In the second step, the reachability graph on the left has

number

number

[0057] In the third step, the reachability graph on the left has

number

[0058] In the fourth step, M2 and M2' are both end identification information.

[0059] As can be seen from the above, M0S0'. The relationship S = {(M0,M0'),(M1,M1'),(M2,M1')),(M3,M2')}.

[0060] Reverse S of S relationship -1 ={(M0′,M0),(M1′,M1),(M1′,M2),(M2′,M3)} is a weak simulation, i.e., the reachable graph to the right of the thick arrow in Figure 5 is weakly simulated by the reachable graph to the left of the black arrow.

[0061] In the first step, the reachability graph on the right contains

number

[0062] In the second step, M1' remains unchanged in the right reachable graph, and M2 exists in the left reachable graph, so

number

[0063] In the third step, the reachability graph on the right has

number

[0064] As can be seen from the above, M0′S -1 M0. From this, we prove that M0 ≒ M0′. Similarly, we can prove that other reachable graphs satisfy weak mutual simulation.

[0065] S303, Accuracy Analysis of Alpha Removal Rule 3: The key issue in performing correctness analysis on α-deletion rule 3 is to prove whether the reachable graph of the source mutual formula MP and the reachable graph of the target mutual formula MP satisfy weak mutual simulation. The proof is as follows.

[0066] That is, we prove that the two reachable graphs shown in Figure 6 satisfy weak mutual simulation.

[0067] In the first step, the reachability graph on the left contains

number

[0068] In the second step, the reachability graph on the left has

number

[0069] In the third step, the reachability graph on the left has

number

[0070] In the fourth step, M3 and M3' are both end identification information.

[0071] As can be seen from the above, M0SM0′. The relationship S={(M0,M0′),(M1,M1′),(M2,M2′)),(M3,M3′)}.

[0072] Reverse S of S relationship -1 ={(M0′,M0),(M1′,M1),(M2′,M2),(M3′,M3)} is a weak simulation, i.e., the reachable graph to the right of the thick arrow in Figure 6 is weakly simulated by the reachable graph to the left of the black arrow.

[0073] In the first step, the reachability graph on the right contains

number

[0074] In the second step, the right reachable graph has either

number

number

[0075] In the third step, the reachability graph on the right has

number

number

[0076] As can be seen from the above, M0′S -1 M0. From here, we prove that M0≒M0′.

[0077] In steps S301, S302, and S303, the accuracy analysis of the α deletion rule requires understanding the concepts of experimental relationship, weak simulation, and weak co-simulation, which are defined as follows:

[0078] Experimental relation: Let Σ = (P, T, F, M0) be a Petri-Net, where M0 is the initial state. The reachable graph of Σ is RG(Σ) = (R(M0), E, ​​Tran), M i ,M j ∈R(M0),s∈Act*, where Act=T∪{α}, and the experimental relation

number

[0079] (1)

number

number

[0080] (2) s = t1...t n in the case of,

number

number

[0081] Weak simulation: If Σ=(P,T,F,M0) is a Petri-Net, its reachable graph is RG(Σ)=(R(M0),E,Tran). If Σ′=(P′,T′,F′,M0′) is a Petri-Net, its reachable graph is RG(Σ′)=(R(,M0′),E′,Tran′), and the relation S is a weak simulation of the reachable information set R(M0)∪R(M0′), and the following condition holds if and only if M0S M0′.

[0082] (1)

number

[0083] (2)

number

[0084] In the case of M0SM0′, Σ is called a Σ′ weak simulation.

[0085] Weak co-simulation: If Σ=(P,T,F,M0) is a Petri-Net, its reachable graph is RG(Σ0=(R(M0),E,Tran). If Σ′=(P′,T′,F′,M0′) is a Petri-Net, its reachable graph is RG(Σ′)=(R(,M0′),E′,Tran′), and a relation B is a weak co-simulation on the reachable information set R(M0)∪R(M0′). If and only if B and its inverse relation are both weak simulations, then the weak co-simulation B satisfies M0SM0′. M0 and M0′ are called weak co-simulation, and M0≒M0′.

[0086] In the case of M0BM0′, Σ and Σ′ satisfy weak cosimulation.

[0087] S4: Initialize the optimized merge process O-MPμ′ as μ′′=(B′,E′,G′,D′,Rd′,Wr′,De′,End′,λ′,M′), and calculate the event set in which data loss errors occur.

number

[0088] The initialization of the optimized merge process is to add condition b′0 and event e′0 before the start condition of μ′, and add condition b′ and event e′ after the end condition, where the b′ and e′ indices are the maximum indices i of the conditions and events in μ′ plus 1, respectively. Also, in μ′′, the stream relations {b′0}×{e′0}, {e′0}×{b0}, {b′ i}×{e′} and {e′}×{b′}. At the same time, e′0 is marked as a delete operation for all data, i.e., De(e′0)=D, and e′ is marked as an end operation, i.e., End(e′)=D.

[0089] S5, analyze the optimized merging process μ′′ after initialization and design a data loss error detection algorithm.

[0090] The data loss error detection algorithm is based on data proximity events. Before calculating data proximity events, a synonymous data operation must be performed on the MP to accurately detect errors even in parallel structures. The synonymous data operation is to provide the data write element d of event e. [e] + If there are two parallel events e1 and e2 belonging to d∈Wr(e1)∪Rd(e1), then add the same data operation label, i.e., (e1,Wr:d or e1,Rd:d), from e1 to e2, where [e] + ={e′∈E|e′≦e}.

[0091] The definition of a data proximity event is as follows. Given two events e1 and e2 such that they satisfy data d ∈ Rd(e1) ∪ Wr(e1) ∪ De(e1), and e1 < e2, if there is an obvious path from e1 to e2 and there is an event e′ on this path that satisfies [Number] or [Number] then e1 is called a data proximity event of e2. The data loss error detection algorithm is as follows.

[0092] [Table 0]

[0093] The algorithm process is as follows. Determine whether event e belongs to the marking function of data element d. If it belongs, in the weak local structure [e] of event e + calculate the data proximity set S(e, d) for the data element d of event e, where [e] + = {e′ ∈ E | e′ ≤ e}. If there is an event e′ of the Wr marking function belonging to the data element d in S(e, d), add event e to the set E_los of events where a data loss error has occurred.

[0094] S6. Output the set of events where data loss has occurred.

[0095] Finally, output the set E_los of events where a data loss error has occurred using the data loss error detection algorithm based on O-MP.

[0096] Some comparison results are shown in Table 1. Currently, proposed data loss error detection methods based on merge processes have not been studied for some time, and therefore cannot detect data loss errors. Furthermore, if a data stream error occurs during transition in reachable graph-based data stream error detection, subsequent transformations may not be triggered, and the related states and arcs may not be displayed in the reachable graph. In this case, the reachable graph does not have a complete state space that overwrites all execution information. Therefore, this method cannot detect all data loss errors. Through comparison, the method of the present invention makes up for this shortcoming. Furthermore, analysis and comparison showed that data loss error detection using the optimized merge process based on the α-deletion rule used in the present invention has better detection results and time than data loss error detection based on conventional merge processes, and is also better than methods based on reachable graphs in terms of state space, detection results, and time.

[0097] Table 1 Comparison of the number of detected data loss errors using MP, O-MP, and reachability graph in CCPS [Table 1]

[0098] The Petri-Net-based information physics system data loss error detection system of the present invention comprises: a Petri net mapping unit for mapping an information-physics system modeled in the business process modeling notation to a first Petri net which is a data-attached Petri net; a merge process generation unit for generating a merge process μ=(B, E, G, D, Rd, Wr, De) of the first Petri net, where B is a condition set, E is an event set, G is an arc set, D is a data element set, Rd is a read marking function, Wr is a write marking function, and De is a delete marking function; A merge process optimization unit for optimizing events and conditions in the merge process μ based on deletion rules to obtain an optimized merge process μ′=(B′,E′,G′,D′,Rd′,Wr′,De′,λ′,M′); In the optimized merge process μ′, it includes a loss error detection unit for determining whether each event is a loss event and outputting a set of events of data loss error; In the loss error detection unit, a synonymous data operation is performed on the optimized merge process μ′, and further data proximity events are calculated. The synonymous data operation is to give the data write element d of event e, [e] + If there are two parallel events e1 and e2 belonging to [e] and d∈Wr(e1)∪Rd(e1), add the same data operation label, that is, (e1,Wr:d or e1,Rd:d) from e1 to e2, provided that [e] + ={e′∈E|e′≦e}, and the data proximity event is to give two events e1 and e2 that satisfy data d∈Rd(e1)∪Wr(e1)∪De(e1). When e1<e2, there is an obvious path from e1 to e2, and the events e′ [[ID=I0]]are

Number

[0099] ​The computer-readable storage medium of the present invention stores a computer program, which, when executed by a processor, realizes the Petri-Net-based information physics system data loss error detection method.

[0100] The computer-readable storage media may include RAM, ROM, EEPROM, CD-ROM or other optical disk storage, magnetic disk storage or other magnetic storage, flash memory, or any other medium that can be used to store desired program code in the form of instructions or data structures and that can be accessed by a computer.

[0101] The processor is used to execute a computer program stored in the memory to implement the various steps of the method according to the above embodiments.

Claims

1. A method for detecting data loss errors in a Petri-Net-based information physics system, comprising: Mapping an information-physics system modeled in the business process modeling notation to a first Petri net that is a Petri net with data; generating a merge process μ=(B, E, G, D, Rd, Wr, De) of the first Petri net, where B is a condition set, E is an event set, G is an arc set, D is a data element set, Rd is a read marking function, Wr is a write marking function, and De is a delete marking function; optimizing the events and conditions in the merging process μ based on a deletion rule to obtain an optimized merging process μ′=(B′, E′, G′, D′, Rd′, Wr′, De′, λ′, M′); In the optimized merging process μ′, determining whether each event is a lost event and outputting a set of events with data loss errors; performing a synonymous data operation on the optimized merging process μ′ and further calculating data proximity events, the synonymous data operation being to provide a data write element d of an event e; [e] ＋ Two parallel events e belonging to １ and e ２ and d∈Wr(e １ ) ∪ Rd(e １ ) in the case of e １ From e ２ to the same data manipulation label, i.e. (e １ , Wr: d or e １ , Rd:d) is added, with the proviso that [e] ＋ = {e′∈E|e′≦e}, and the data proximity event is a data d∈Rd(e １ ) ∪ Wr(e １ ) ∪ De(e １ ) two events e １ and e ２ is to give e １ <e ２ In the case of e １ From e ２ There is clearly a path to [Number 48] In the case of e １ wo e ２ and calling the data proximity event The method for determining whether each event is a lost event is as follows: Determine whether event e belongs to the marking function of data element d, and if so, determine the weak local structure [e] of event e. ＋ In the above, a data proximity set S(e, d) for an event e and a data element d is calculated; if S(e, d) contains an event e′ that belongs to the marking function of the data element d, the event e is added to an event set E_los in which a data loss error occurs; The merging process μ is a target mutual merging process μ λ = (B, E, G, D, Rd, Wr, De, λ, M), where λ is a marking function used to mark the type of the current event, M:B → {0, 1, 2, 3, ...} is the identity of the net, and the marking function is λ:E → L ∪ {α}, where L = {!, ?, @} × D includes non-marking α, write !, read ?, delete @; The target mutual merging process μ λ In ,delete the rule deletion events and conditions to obtain the optimized merging process μ′, The deletion rule is: ∃e∈E∧λ(e)=α, [Number 49] This event e can be deleted if [Number 50] D′=D, Rd′=Rd, Wr′=Wr, De′=De, λ′=λ−{(e,α)}, [Equation 51] And, ∃e １ ,e ２ ,... e ｎ ∈E∧λ(e １ ) = α ∧ λ(e ２ )=α∧. .. .. ∧λ(e ｎ )=α, [Number 52] In the case of １ , e ２ ,... e ｎ can be deleted, and we obtain a new MPμ′=(B′,E′,G′,D′,Rd′,Wr′,De′,λ′,M′), where [Number 53] D′=D, Rd′=Rd, Wr′=Wr, De′=De, λ′=λ-{(e １ ,a)}-{(u ２ ,a)...-{(u ｎ ,a)}, [Number 54] where the value of i is 1 to n; ∃e １ ,e ２ ,... e ｎ ∈E∧λ(e １ ) = α ∧ λ(e ２ )=α∧. .. .. ∧λ(e ｎ ) = α ∧ λ(e) ≠ α, [Number 55] All α in can be deleted, and e events cannot be deleted. Add α to maintain the collision structure in the deletion rules, with the exception that B'=B, [Number 56] D′=D, Rd′=Rd, Wr′=Wr, De′=De, λ′=λ, and M'=M.

2. The step of mapping the information physics system modeled in the business process modeling notation to the first Petri net, which is a Petri net with data, includes: mapping the information-physics system to a first Petri net based on data objects and states in the information-physics system and modeling the first Petri net; The first Petri net has an initial marker m ０ Net N with data Ｄ and Σ=(N Ｄ , m ０ ), N Ｄ = (P, T, F, D, Read, Write, Delete), where (P, T, F) is a net, P is a library, T is a transition, F is a stream relation, and Read, Write, and Delete are respectively a read marking function, a write marking function, and a delete marking function in the first Petri net.

3. The step of mapping the information physics system modeled in the business process modeling notation to the first Petri net, which is a Petri net with data, includes: The merge process μ=Merge(β) is included, and the method for generating the merge process μ is as follows: Merging all β-conflict conditions with the same label and occurrence depth into one condition; Merge all β events with the same label, i.e., the before and after sets, into one event. All related arcs of β must be redirected based on the merge conditions and events, 3. The method for detecting data loss errors in an information physics system based on Petri-Net as claimed in claim 2, wherein β is the branching process of the first Petri net.

4. The step of mapping the information physics system modeled in the business process modeling notation to the first Petri net, which is a Petri net with data, includes: Mapping exclusive gateways of the business process modeling notation to a conflict structure of the first Petri net; mapping parallel gateways of the business process modeling notation to the parallel structure of the first Petri net; Mapping activities to transitions and their pre- or post-libraries; Mapping inputs and outputs of a data object or data library to marking functions of the data read / write elements during the transition; Mapping an external input to a marking function of the data writing element during the transition; Mapping message streams between activities to libraries and their pre- and post-transitions; and mapping other connection objects to arcs between related libraries and transitions.

5. The step of optimizing the events and conditions in the merging process μ based on the deletion rule to obtain an optimized merging process μ′=(B′, E′, G′, D′, Rd′, Wr′, De′, λ′, M′) is as follows:

2. The method for detecting data loss errors in an information physics system based on Petri-Net according to claim 1, further comprising the steps of: performing an accuracy analysis on the deletion rule; adopting weak reciprocity as a criterion for detecting whether behaviors are consistent; and reducing the problem of detecting whether the source merge process and the target reciprocal merge process maintain behavior consistency to the problem of detecting whether the source merge process and the target reciprocal merge process satisfy weak reciprocity.

6. In the optimized merging process μ′, the step of determining whether each event is a loss event and outputting a set of events of data loss errors includes: In the optimized merging process μ′, before determining whether each event is a lost event, the optimized merging process μ′ is initialized, and the initialization includes: The condition b' is before the starting condition of the optimized merging process μ'. ０ and event e' ０ and add a condition b′ and an event e′ after the termination condition, where the b′ and e′ subscripts are the highest subscript i of the conditions and events in μ′ plus 1, respectively; In the optimized merge process μ′, the stream relation {b′ ０ }×{e′ ０ }, {e′ ０ }×{b ０ }, {b' ｉ }×{e′} and {e′}×{b′}; e' ０ Mark as a deletion operation for all data, i.e., De(e' ０ )=D, and marking e' as a termination operation, i.e., End(e')=D, where End is a termination marking function.

7. A data loss error detection system in a Petri-Net-based information physics system, comprising: a Petri net mapping unit for mapping an information physics system modeled in the business process modeling notation to a first Petri net which is a Petri net with data; a merge process generation unit for generating a merge process μ=(B, E, G, D, Rd, Wr, De) of the first Petri net, where B is a condition set, E is an event set, G is an arc set, D is a data element set, Rd is a read marking function, Wr is a write marking function, and De is a delete marking function; a merge process optimization unit for optimizing events and conditions in the merge process μ based on a deletion rule to obtain an optimized merge process μ′=(B′, E′, G′, D′, Rd′, Wr′, De′, λ′, M′); In the optimized merging process μ′, a loss error detection unit is included for determining whether each event is a loss event and outputting a set of data loss error events; In the loss error detection unit, a synonymous data operation is performed on the optimized merging process μ′, and a data proximity event is further calculated, and the synonymous data operation is to provide a data write element d of the event e, [e] ＋ Two parallel events e belonging to １ and e ２ and d∈Wr(e １ ) ∪ Rd(e １ ) in the case of e １ From e ２ to the same data manipulation label, i.e. (e １ , Wr: d or e １ , Rd:d) is added, with the proviso that [e] ＋ = {e′∈E|e′≦e}, The data proximity event is a data d∈Rd(e １ ) ∪ Wr(e １ ) ∪ De(e １ ) two events e １ and e ２ is to give e １ <e ２ In the case of e １ From e ２ There is clearly a path to [Number 57] In the case of e １ wo e ２ is called a data proximity event, The method for determining whether each event is a lost event is as follows: Determine whether event e belongs to the marking function of data element d, and if so, determine the weak local structure [e] of event e. ＋ In step S200, a data proximity set S(e, d) for an event e and a data element d is calculated. If S(e, d) contains an event e′ that belongs to the marking function of the data element d, the event e is added to an event set E_los in which a data loss error has occurred. The merging process μ is a target mutual merging process μ λ = (B, E, G, D, Rd, Wr, De, λ, M), where λ is a marking function used to mark the type of the current event, M:B → {0, 1, 2, 3, ...} is the identity of the net, and the marking function λ:E → L ∪ {α}, where L = {!, ?, @} × D includes non-marking α, write !, read ?, delete @, The target mutual merging process μ λ In ,delete the rule deletion events and conditions to obtain the optimized merging process μ′, The deletion rule is: ∃e∈E∧λ(e)=α, [Number 58] This event e can be deleted if [Number 59] D′=D, Rd′=Rd, Wr′=Wr, De′=De, λ′=λ−{(e,α)}, [Number 60] And, ∃e １ ,e ２ ,... e ｎ ∈E∧λ(e １ ) = α ∧ λ(e ２ )=α∧. .. .. ∧λ(e ｎ )=α, [Number 61] In the case of １ , e ２ ,... e ｎ can be deleted, and we obtain a new MPμ′=(B′,E′,G′,D′,Rd′,Wr′,De′,λ′,M′), where [Number 62] D′=D, Rd′=Rd, Wr′=Wr, De′=De, λ′=λ-{(e １ ,a)}-{(u ２ ,a)...-{(u ｎ ,a)}, [Number 63] where the value of i is 1 to n; ∃e １ ,e ２ ,... e ｎ ∈E∧λ(e １ ) = α ∧ λ(e ２ )=α∧. .. .. ∧λ(e ｎ ) = α ∧ λ(e) ≠ α, [Number 64] All α in can be deleted, and e events cannot be deleted. Add α to maintain the collision structure in the deletion rules, with the exception that B'=B, [Number 65] D′=D, Rd′=Rd, Wr′=Wr, De′=De, λ′=λ, and M'=M.

8. A computer-readable storage medium on which a computer program is stored, A computer-readable storage medium, characterized in that, when the computer program is executed by a processor, it realizes the method for detecting data loss errors in an information physics system based on Petri-Net according to any one of claims 1 to 6.

Images