A manufacturing big data quality identification method based on aging-dependent rules

Abstract

The application provides a manufacturing big data quality identification method based on time-dependent rules, and relates to the technical field of data quality identification.The method comprises the following steps: determining the non-deterministic dependency relationship of a process according to the non-deterministic dependency relationship of data conclusions, constructing a weighted time graph based on a target time-dependent rule, and determining the timeliness of current data at a target time according to the weighted time graph; comparing the absolute value of a first difference between the timeliness of current data at the target time and a preset timeliness identification lower limit with the absolute value of a second difference between the timeliness of current data at the target time and a preset timeliness identification upper limit; in the case that the absolute value of the first difference is not greater than the absolute value of the second difference, identifying that the timeliness of current data is invalid; in the case that the absolute value of the first difference is greater than the absolute value of the second difference, determining that the timeliness of current data is valid, and finally effectively identifying the quality of manufacturing big data from the timeliness dimension.

Description

Technical Field

[0001] This invention relates to the field of data quality assessment technology, and in particular to a method for assessing the quality of manufacturing big data based on time-dependent rules. Background Technology

[0002] As manufacturing becomes increasingly intelligent, numerous sensors will collect various types of interdependent data during the production process. Effectively assessing the quality of this interdependent data is crucial for enterprise production. In actual production, there are many dimensions for assessing data quality, among which the assessment of data timeliness is a significant aspect of data quality research.

[0003] However, in real-world industrial databases, timestamps are often ambiguous. Different attributes of the same record change at different frequencies over time, resulting in varying degrees of newness for these attributes. Furthermore, storing a corresponding timestamp for each data item in every process is not easy. Current technologies based on timeliness rules, without relying on timestamps, can only identify the temporal relationship of data items, but cannot determine their timeliness.

[0004] Therefore, a new method for timeliness assessment is urgently needed. Summary of the Invention

[0005] This invention provides a method for quality assessment of manufacturing big data based on time-dependent rules, in order to at least partially solve the problems existing in related technologies.

[0006] The first aspect of this invention provides a method for quality assessment of manufacturing big data based on time-dependent rules, the method comprising:

[0007] Based on the conclusions between every two data points in the acquired dataset and the nondeterministic dependencies between each conclusion, a conclusion dependency sequence graph is constructed, wherein each conclusion is taken as a conclusion node in the conclusion dependency sequence graph, and directed edges represent the nondeterministic dependencies between each conclusion node.

[0008] The conclusion nodes containing the same process in the conclusion dependency sequence graph are merged to construct a process dependency sequence graph. The process corresponding to the merged conclusion node is used as the process node of the process dependency sequence graph, and the non-deterministic dependency relationship between each process node is represented by a directed edge.

[0009] Based on the target timeliness dependency rule, the target time in the target timeliness dependency rule is used as the initial node, and each data in the target process is used as the intermediate node to construct the weighted timeliness diagram of the target process;

[0010] Based on the weighted timeliness diagram, determine the timeliness of the current data in the target process at the target time;

[0011] Compare the absolute values ​​of the first difference between the timeliness of the current data at the target time and the second difference between the preset lower limit of the timeliness assessment value and the preset upper limit of the timeliness assessment value;

[0012] If the absolute value of the first difference is not greater than the absolute value of the second difference, the timeliness of the current data is determined to be invalid.

[0013] If the absolute value of the first difference is greater than the absolute value of the second difference, the timeliness of the current data is determined to be valid.

[0014] Optionally, determining the timeliness of the current data in the target process at the target time based on the weighted timeliness diagram includes:

[0015] The timeliness of the current data at the target time is calculated using the following formula:

[0016] tml(t[A]) = tml(t < A τ)

[0017] Where tml(t[A]) represents the timeliness of the current data at the target time; t < A τ represents the conclusion that data t precedes the target time τ in process A; tml(t < A τ) represents the dependence strength of the conclusion that the data t under process A is valid before the target time τ; t is the current data; A is the process; τ is the target time.

[0018] Optionally, a tuple may include multiple steps. Considering a tuple, determining the timeliness of the current tuple at the target time includes:

[0019] The timeliness of the current tuple at the target time is calculated using the following formula:

[0020]

[0021] Where d is the current tuple, and Ttml(d) represents the timeliness of the current tuple d; w i This represents the distinct data t in the current tuple d. i Their respective weights; tml(t[A i ]) indicates the timeliness of different processes.

[0022] Optionally, a dataset may consist of multiple tuples. When considering a dataset, determining the timeliness of the current dataset includes:

[0023] The timeliness of the current dataset is calculated using the following formula:

[0024]

[0025] Where D is the current dataset, and Atml(D) represents the timeliness of the current dataset D; W i Represents distinct tuples d in the current dataset D. i Their respective weights; tml(d i () indicates the timeliness of different tuples.

[0026] Optionally, the strength of the dependency on the conclusion that the current data is valid before the target time is determined, including:

[0027] The dependence strength of the conclusion that the current data in the target process is valid before the target time is calculated using the following formula:

[0028] tml(rhs(r))=tml(lhs(r))×tml(r)

[0029] Where r is the target time-dependent rule; rhs(r) means that under the target time-dependent rule r, rhs(r) represents the conclusion t < A τ; tml(rhs(r)) represents the conclusion t < A τ represents the dependency strength under the target time-dependent rule r; lhs(r) represents the left-hand side conclusion of the target time-dependent rule r under the target pre-defined time-dependent rule r; tml(lhs(r)) represents the dependency strength under the target time-dependent rule r; tml(r) represents the dependency strength of the target time-dependent rule r.

[0030] In the case of multiple target timeliness dependency rules, the maximum value among the multiple dependency strengths of the conclusion that the current data is valid before the target time, calculated based on the multiple target timeliness dependency rules, is determined as the dependency strength of the conclusion that the current data is valid before the target time.

[0031] Optionally, determining the timeliness of the current data under the target process at the target time based on the weighted timeliness chart includes:

[0032] Based on the nondeterministic dependency relationship between the target time and the target data in the current data in the weighted timeliness diagram, the flux value of the target time and the target data is determined as the timeliness of the target data at the target time; the flux value characterizes the strength of the nondeterministic dependency relationship between the target time and the target data.

[0033] In the case where only one nondeterministic dependency exists, the minimum throughput value between the target time and the target data is determined as the timeliness of the target data at the target time; or,

[0034] In the presence of at least two nondeterministic dependencies, the maximum value among the throughput values ​​of the target time and the target data under at least two nondeterministic dependencies is determined as the timeliness between the target time and the target data.

[0035] Optionally, the step of constructing a weighted timeliness graph of the target process based on the target timeliness dependency rule, using the target time in the target timeliness dependency rule as the initial node and each data under the target process as the intermediate node, specifically includes the following steps:

[0036] S1. Obtain a set of time-dependent rules, and determine a first target time-dependent rule containing the target time from the set of time-dependent rules;

[0037] S2. Based on the set of time-dependent rules, determine the second target time-dependent rule;

[0038] S3. Take the target time in the first target timeliness dependency rule as the initial node, and take each piece of data in the target process that can satisfy the second target timeliness dependency rule as the intermediate node;

[0039] S4. Add a directed edge between any two nodes with nondeterministic dependencies among the initial node and the intermediate nodes to construct a weighted time graph of the target process.

[0040] Optionally, the manufacturing big data quality assessment method based on timeliness-dependent rules according to claim 1 is characterized in that, based on the weighted timeliness diagram, determining the timeliness of each data point under the target process includes:

[0041] Based on the nondeterministic dependency relationship between any two data points in each data point under the target process, the conclusions corresponding to each of the two data points are obtained.

[0042] Based on the target timeliness dependency rule, calculate the dependency strength of the conclusions corresponding to any two of the data;

[0043] The timeliness of each piece of data under the target process is determined based on the strength of dependence of the conclusions corresponding to any two of the data.

[0044] Optionally, the target timeliness dependency rules include: a first dependency rule, a second dependency rule, a third dependency rule, a fourth dependency rule, a fifth dependency rule, a sixth dependency rule, and a seventh dependency rule; obtaining the conclusion between every two data points in the dataset includes:

[0045] Based on the first dependency rule, the second dependency rule, the third dependency rule, the fourth dependency rule, the fifth dependency rule, the sixth dependency rule, and the seventh dependency rule, determine the explicit nondeterministic dependency relationship between every two data points in the dataset.

[0046] The explicit nondeterministic dependency between any two data points in the dataset is taken as the conclusion between any two data points in the dataset.

[0047] Optionally, the second target timeliness dependency rule further includes: an eighth dependency rule; the method further includes:

[0048] Based on the eighth dependency rule, the implicit nondeterministic dependencies between every two data points in the dataset are determined.

[0049] The present invention has the following advantages:

[0050] This invention proposes a manufacturing big data quality assessment method based on timeliness dependency rules. This method determines the non-deterministic dependencies between different processes based on the non-deterministic dependencies between conclusions of various data points. Then, based on target timeliness dependency rules and the non-deterministic dependencies between processes, a weighted timeliness graph is constructed. The timeliness of the current data at the target time is determined based on this weighted timeliness graph. The absolute values ​​of the differences between the current data's timeliness at the target time and a first value (a lower limit of the preset timeliness assessment value) and a second value (a higher limit of the preset timeliness assessment value) are compared. If the absolute value of the first difference is not greater than the absolute value of the second difference, the current data is deemed invalid. If the absolute value of the first difference is greater than the absolute value of the second difference, the current data is deemed valid. This manufacturing big data quality assessment method based on timeliness dependency rules overcomes the limitations of timestamps on data quality assessment. It can determine the timeliness between data points solely based on data timeliness dependency rules even in the absence of timestamps, and ultimately effectively assess the quality of manufacturing big data from the perspective of timeliness. Attached Figure Description

[0051] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the description of the embodiments of the present invention will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0052] Figure 1 This is a flowchart of a manufacturing big data quality assessment method based on time-dependent rules, according to an embodiment of the present invention.

[0053] Figure 2 This is a schematic diagram illustrating the determination of path throughput value according to an embodiment of the present invention;

[0054] Figure 3 This is a schematic diagram of a conclusion-dependency sequence diagram according to an embodiment of the present invention;

[0055] Figure 4 This is a schematic diagram of merging conclusion dependency sequence diagrams according to an embodiment of the present invention;

[0056] Figure 5 This is a flowchart of a data quality assessment method according to an embodiment of the present invention. Detailed Implementation

[0057] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0058] In actual production and manufacturing processes, each step generates a large amount of interdependent data. For enterprises, effectively assessing the quality of this interdependent data is crucial. Data quality assessment involves multiple dimensions, among which timeliness is one of the important dimensions in the field of data quality research, especially in the industrial manufacturing sector where the timeliness of data from each step is of particular importance.

[0059] However, in real-world industrial databases, the problem of ambiguous timestamps frequently exists. That is, different attributes of the same record change at different frequencies over time, resulting in varying degrees of newness for different attributes within the same record. Since storing a corresponding timestamp for each data item is not easy, a method that does not rely entirely on timestamps is needed to determine the timeliness of data.

[0060] Timeliness assessment can be divided into qualitative and quantitative assessments. Qualitative assessment refers to a binary state such as "outdated" or "not outdated," while quantitative assessment refers to the likelihood of a data item or data set becoming outdated. For example, a data set instance may consist of multiple tuples, and each tuple may contain multiple data items. In actual production, a data set can be a batch of workpieces or workpieces of the same type; each tuple can be data for a workpiece in each process, an attribute of the workpiece itself, or both process data and workpiece attributes. Based on the manufacturing big data quality assessment method based on timeliness-dependent rules proposed in this application, the timeliness of each data item, each tuple, and the data set can be quantitatively assessed.

[0061] Reference Figure 1 The diagram illustrates a flowchart of a manufacturing big data quality assessment method based on time-dependent rules according to an embodiment of the present invention. The manufacturing big data quality assessment method based on time-dependent rules provided by this embodiment of the present invention may include the following steps:

[0062] Step S101: Based on the conclusions between every two data points in the acquired dataset and the nondeterministic dependencies between each conclusion, a conclusion dependency sequence graph is constructed, wherein each conclusion is taken as a conclusion node of the conclusion dependency sequence graph, and a directed edge represents the nondeterministic dependency between each conclusion node.

[0063] Specifically, a dataset is retrieved from an industrial database. This dataset contains multiple data points. Data analysis is performed on these data points to determine the relationships between any two data points, and further, to determine whether the dependencies between these dependencies are nondeterministic. In practical applications, determining whether the dependencies between dependencies are nondeterministic requires identifying whether each dependency is deterministic or nondeterministic. If it is deterministic, it takes the form t... i [A]op t j [A] or t k If the conclusion of [A]op a is a deterministic conclusion, where A is the target process, a is a constant, and op∈{=,≠,>,<,≥,≤}, then it belongs to a deterministic dependency relationship; if it is a non-deterministic conclusion, such as t i < A t j or t i < A If the conclusion of τ is nondeterministic, then it belongs to a nondeterministic dependency relationship. After determining the nondeterministic dependencies between conclusions, we construct a conclusion dependency sequence graph by using each conclusion as a conclusion node and directed edges to represent the nondeterministic dependencies between each conclusion node.

[0064] In practical applications, suppose there is a process A containing multiple data points. If data analysis of process A determines that the dependencies between the conclusions of these multiple data points are nondeterministic, then the data in process A will be used to construct two types of conclusion nodes: (<, A, *), and (<, A, τ). Here, (<, A, *), represents the relationship between t and τ. i < A t j The conclusion of this type is characterized in process A, t i Prior to t j The conclusion is that (<, A, τ) is used to represent t. i < A A conclusion of type τ, characterized in process A, t i The conclusion precedes time τ.

[0065] Furthermore, by scanning the preset set of time-dependent rules, it is determined whether there exists at least one time-dependent rule in the set that makes conclusion t... i < A τ and conclusion t i < A t j If the time-dependent rule is satisfied, a directed edge is added between the two conclusions, thus constructing a conclusion dependency sequence graph. The directed edge is used to represent the non-deterministic dependency relationship between the two conclusions.

[0066] It is important to note that in a conclusion dependency order graph, if three conclusions influence each other, the graph alone cannot uniquely determine the order in which the conclusions are judged in terms of their timeliness. Furthermore, it is clear that the strength of the dependency between conclusions is often difficult to obtain when cyclical rules exist. For example, suppose the following three rules exist in the rule set: and The three rules influence each other; that is, there is a cyclic dependency among them. Therefore, we cannot derive tml(e i < A e j ), tml(e i < B e j ) and tml(e i < C e j Therefore, it is necessary to further optimize the conclusion dependency sequence graph to eliminate the dependency relationship of loops and obtain a reasonable time-sensitive calculation order among the conclusions.

[0067] Step S102: Merge the conclusion nodes that contain the same process in the conclusion dependency sequence graph to construct a process dependency sequence graph. The process corresponding to the merged conclusion node is taken as the process node of the process dependency sequence graph, and the non-deterministic dependency relationship between each process node is represented by a directed edge.

[0068] Specifically, among the nondeterministic dependencies between conclusions, there are both explicit and implicit nondeterministic dependencies. In the steps described above, based on the constructed conclusion dependency sequence diagram, the explicit nondeterministic dependencies between conclusions can be determined, but the implicit nondeterministic dependencies cannot be determined. Therefore, based on the conclusion dependency sequence diagram, it is necessary to determine the process corresponding to each conclusion node in the conclusion dependency sequence diagram, and merge the conclusion nodes with the same process in the process corresponding to each conclusion node, so as to construct the process dependency sequence diagram.

[0069] For example: when determining the conclusion depends on the conclusion node t in the sequence diagram containing the same process. i < A t j and t i < A When dealing with two types of nondeterministic conclusions (τ), the nodes representing these two types of conclusions are transformed into process nodes (*, <, A, *) and (*, <, A, τ) corresponding to process A. Directed edges represent the dependencies between process nodes with nondeterministic dependencies, thus transforming the conclusion dependency sequence graph into a process dependency sequence graph. Then, the transformed process dependency sequence graph is traversed. If multiple identical dependencies exist, i.e., multiple directed edges exist between two process nodes, these multiple directed edges are merged into a single directed edge. It is important to note that in the transformed process dependency sequence graph, directed edges no longer represent relationships between conclusions, but rather relationships between processes.

[0070] In practical applications, if the length of the directed edge between two process nodes is 1, it indicates that the two processes have a direct dependency relationship. If the length of the directed edge between two process nodes is greater than or equal to 2, it indicates that the two processes have an indirect dependency relationship.

[0071] Furthermore, after merging conclusion nodes that contain the same process in the conclusion dependency sequence graph to construct the process dependency sequence graph, it is necessary to perform topological sorting on the process dependency sequence graph to determine the calculation order of the timeliness of each process in the process dependency sequence graph.

[0072] Specifically, topological sorting of the process dependency sequence graph can yield a total order of attributes based on dominance relationships. Specifically, the initial sort is set to empty by default. Each time, a node with an in-degree of zero is selected, and its corresponding process node is added to the current sort. Then, the node and its corresponding edge are deleted. After that, another node with an in-degree of zero is selected, and its corresponding attribute is added to the current sort. This process is repeated until the process dependency sequence graph becomes an empty graph, that is, the topological sorting is completed. At this point, the reasonable calculation order of the timeliness of each process in the process dependency sequence graph can be determined.

[0073] It is important to note that the above process dependency sequence graph does not contain loops, because in actual production, there should be no dependency loops between processes. Furthermore, if the process dependency graph merged through the conclusion dependency graph contains loops, it is not applicable to the minimum feedback arc set problem, because the dependencies between processes have practical production significance and cannot be simply removed by removing a few edges.

[0074] Step S103: Based on the target timeliness dependency rule, take the target time in the target timeliness dependency rule as the initial node and take each data in the target process as the intermediate node to construct the weighted timeliness diagram of the target process;

[0075] In step S102, after topological sorting of the process dependency sequence diagram to determine the reasonable calculation order of timeliness between each process, in this step, the target timeliness dependency rule is determined from the timeliness dependency rule set, and the target time in the target timeliness dependency rule is used as the initial node, and each data in the target process is used as the intermediate node to construct the weighted timeliness diagram corresponding to the target process.

[0076] In this embodiment of the invention, the step of constructing a weighted timeliness graph of the target process based on the target timeliness dependency rule, using the target time in the target timeliness dependency rule as the initial node and each data under the target process as the intermediate node, specifically includes the following steps:

[0077] S1. Obtain a set of time-dependent rules, and determine a first target time-dependent rule containing the target time from the set of time-dependent rules;

[0078] S2. Based on the set of time-dependent rules, determine the second target time-dependent rule;

[0079] S3. Take the target time in the first target timeliness dependency rule as the initial node, and take each piece of data in the target process that can satisfy the second target timeliness dependency rule as the intermediate node;

[0080] S4. Add a directed edge between any two nodes with nondeterministic dependencies among the initial node and the intermediate nodes to construct a weighted time graph of the target process.

[0081] Step S104: Determine the timeliness of the current data in the target process at the target time based on the weighted timeliness diagram;

[0082] Specifically, it is necessary to identify the target process requiring timeliness calculation from the current processes, and use each data point in the target process as an intermediate node in constructing the weighted timeliness graph of that target process; secondly, iterate through the set of timeliness dependency rules to determine the target timeliness dependency rule, which is a rule whose right-hand side is of the form: t i < A t j or t k < A τ k The conclusion of this type has a time-dependent rule, and the target time-dependent rule contains the target time τ. k .

[0083] In this embodiment of the application, the weighted time-efficiency diagram is further defined as follows:

[0084] (a).WTG A It contains two types of nodes, namely, those representing t respectively. i < A t j and t i < A The data items t and time τ in these two types of conclusions are both based on a certain time τ as the initial node of the WTG graph, where τ includes the original time in the dataset and the threshold θ that we set.

[0085] (b) For each rule r in set Σ, if there exists τ∈T∪{θ} such that... If the right-hand side of the rule contains a conclusion at a given time, then it is defined as the initial node of the WTG graph.

[0086] (c) For each rule r in set Σ, if t1, t2 ∈ D such that rhs(r) = t1 < A If t2, tml(lhs(r))>0 and tml(r)>0, then t1< A t2 can be directly derived from r. To represent this time relationship, a directed edge t2→t1 is added from t2 to t1. If this directed edge already exists, its weight is updated.

[0087] (d) The weight of the directed edge (t1, t2) in the WTG graph is denoted as weight(t1, t2) = tml(r) × tml(lhs(r)). If there exist k rules {r1, r2, <, r k} can all directly deduce t1 < A At time t2, we will use the larger value as the timeliness of this conclusion, where weight(t1,t2) = max{tml(r1)×tml(lhs(r1)),tml(r2)×tml(lhs(r2))},…,tml(r k )×tml(lhs(r k )).

[0088] Where D represents the set of data, Σ represents the set of time-dependent rules, T represents the set of all time-dependent rules in Σ containing time points, A represents a certain process, θ is the effective time threshold of D, and the weighted time-dependent graph of process A is denoted as WTG. A .

[0089] In a preferred embodiment, determining the timeliness of the current data under the target process at the target time based on the weighted timeliness chart includes:

[0090] Based on the nondeterministic dependency relationship between the target time and the target data in the current data in the weighted timeliness diagram, the flux value between the target time and the target data is determined as the timeliness of the target data at the target time; the flux value characterizes the strength of the nondeterministic dependency relationship between the target time and the target data; wherein, when only one nondeterministic dependency relationship exists, the minimum flux value between the target time and the target data is determined as the timeliness of the target data at the target time; or, when at least two nondeterministic dependencies exist, the maximum value among the flux values ​​between the target time and the target data under at least two nondeterministic dependencies is determined as the timeliness between the target time and the target data.

[0091] Specifically, the target time-dependent rules are determined from the acquired set of time-dependent rules. The target time in these rules is used as the initial node for constructing the weighted time-dependent graph, and the current data under the target process is used as the intermediate node. Directed edges represent the non-deterministic dependencies between the data, thus constructing the weighted time-dependent graph. Based on the target time-dependent rules, the flux value between the target time and the target data in the current data is calculated. This flux value represents the strength of the non-deterministic dependency between the target time and the target data, and can also be referred to as the path-to-path flux between the node corresponding to the target time and the node corresponding to the target data in the weighted time-dependent graph.

[0092] Among them, there is only one nondeterministic dependency between the target time and the target data, that is, there is only one path. The minimum throughput value between the target time and the target data is determined as the timeliness of the target data at the target time. In other words, the minimum path single throughput between the node corresponding to the target time and the node corresponding to the target data is determined as the timeliness of the target data at the target time.

[0093] Alternatively, if there are at least two nondeterministic dependencies between the target time and the target data, i.e., at least two paths exist, the maximum value of the flux between the target time and the target data is determined as the timeliness of the target data at the target time. In other words, the maximum single-path flux among the multiple single-path fluxes between the node corresponding to the target time and the node corresponding to the target data is determined as the timeliness of the target data at the target time.

[0094] Reference Figure 2 The diagram shown is a schematic representation of a path flux value provided by the present invention. Figure 2 As can be seen, there are multiple nondeterministic dependencies between the target time τ and the target data t4, i.e., there are multiple paths. Assume that... Figure 2 In the given condition, when only the path τ→t1→t2→t4 exists, the flux value of this path is:

[0095] Flux(Path1(τ,t4))=min{weight(t4< A t2)=0.9, weight(t2< A t1)=0.8, weight(t1< A τ=1)}=0.8, that is, the minimum value of the flux value in this path is determined as the timeliness of the target data t4 at the target time τ.

[0096] like Figure 2 In the case where there are three paths τ→t1→t2→t4, τ→t4, and τ→t3→t4, sFlux(Path(τ,t4))=Max{Flux(Path1(τ,t4))=0.8,Flux(Path2(τ,t4))=0.6,Flux(Path3(τ,t4))=0.7}=0.8, that is, the maximum value of the flux value among the three paths is determined as the timeliness of the target data t4 at the target time τ.

[0097] In a preferred embodiment, determining the timeliness of the current data in the target process at the target time based on the weighted timeliness diagram includes:

[0098] The timeliness of the current data at the target time is calculated using the following formula:

[0099] tml(t[A]) = tml(t < A τ) (1)

[0100] Where tml(t[A]) represents the timeliness of the current data at the target time; t < A τ represents the conclusion that data t occurs before the target time τ in process A. tml(t < t) A τ) represents the dependence strength of the conclusion that the data t under process A is valid before the target time τ; t is the current data; A is the process; τ is the target time.

[0101] Specifically, when it is necessary to determine the timeliness of the current data in the target process at the target time, that is, to determine the strength of the dependence of the conclusion that the current data was valid before the target time.

[0102] It is important to note that the strength of the dependency on the conclusion that the current data was valid before the target time can be calculated using the following formula:

[0103] tml(rhs(r))=tml(lhs(r))×tml(r) (2)

[0104] Where r is the target time-dependent rule; rhs(r) means that under the target time-dependent rule r, rhs(r) represents the conclusion t < A τ; tml(rhs(r)) represents the conclusion t < A τ represents the dependency strength under the target time-dependent rule r; lhs(r) represents the left-hand side conclusion of the target time-dependent rule r under the target pre-defined time-dependent rule r; tml(lhs(r)) represents the dependency strength under the target time-dependent rule r; tml(r) represents the dependency strength of the target time-dependent rule r.

[0105] In the case of multiple target timeliness dependency rules, the maximum value among the multiple dependency strengths of the conclusion that the current data is valid before the target time, calculated based on the multiple target timeliness dependency rules, is determined as the dependency strength of the conclusion that the current data is valid before the target time.

[0106] In a preferred embodiment, a tuple is a data set containing multiple processes. In the above embodiments, the timeliness of the current data in the target process was determined. In this embodiment, considering a tuple, determining the timeliness of the current tuple at the target time includes:

[0107] The timeliness of the current tuple at the target time is calculated using the following formula:

[0108]

[0109] Where d is the current tuple, and Ttml(d) represents the timeliness of the current tuple d; w i This represents the distinct data t in the current tuple d. i Their respective weights; tml(t[A i The tuple represents the timeliness of different processes. That is, the timeliness of a tuple is the weighted sum and average of the timeliness of each process.

[0110] In a preferred embodiment, a dataset comprises multiple tuples. Considering only one dataset, the timeliness of the current data is determined by calculating the timeliness of the current dataset using the following formula:

[0111]

[0112] Where D is the current dataset, and Atml(D) represents the timeliness of the current dataset D; W i Represents distinct tuples d in the current dataset D. i Their respective weights; tml(d i The '(')' indicates the timeliness of different tuples. In practical applications, each tuple can represent the same product from different batches, or different products from the same batch. When each tuple represents the same product from different batches, the weight of the tuple represents the different accuracy requirements of different batches; when each tuple represents different products from the same batch, the weight of the tuple often takes the same value.

[0113] Step S105: Compare the absolute values ​​of the first difference between the timeliness of the current data at the target time and the second difference between the preset lower limit of the timeliness assessment value and the preset upper limit of the timeliness assessment value.

[0114] Step S106: If the absolute value of the first difference is not greater than the absolute value of the second difference, the timeliness of the current data is determined to be invalid.

[0115] Step S107: If the absolute value of the first difference is greater than the absolute value of the second difference, the timeliness of the current data is determined to be valid.

[0116] Specifically, after calculating the timeliness of the current data, it is necessary to further verify whether the timeliness of the current data is valid.

[0117] In this embodiment of the invention, an upper limit and a lower limit for timeliness assessment are preset. The absolute value of the first difference between the timeliness of the current data at the target time and the preset lower limit is compared with the absolute value of the second difference between the current data and the preset upper limit. If the absolute value of the first difference is not greater than the absolute value of the second difference, the current data is deemed invalid. If the absolute value of the first difference is greater than the absolute value of the second difference, the current data is deemed valid.

[0118] In a preferred embodiment, the target timeliness dependency rule includes: a first dependency rule, a second dependency rule, a third dependency rule, a fourth dependency rule, a fifth dependency rule, a sixth dependency rule, and a seventh dependency rule; obtaining the conclusion between every two data points in the dataset includes:

[0119] Based on the first dependency rule, the second dependency rule, the third dependency rule, the fourth dependency rule, the fifth dependency rule, the sixth dependency rule, and the seventh dependency rule, determine the explicit nondeterministic dependency relationship between every two data points in the dataset.

[0120] The explicit nondeterministic dependency between any two data points in the dataset is taken as the conclusion between any two data points in the dataset.

[0121] In this embodiment, the second target timeliness dependency rule further includes: an eighth dependency rule; the method further includes:

[0122] Based on the eighth dependency rule, the implicit nondeterministic dependencies between every two data points in the dataset are determined.

[0123] For example, Table 1 shows a process flow example provided by the present invention, and Table 2 shows the preset target timeliness dependency rules. The process contains two tuples, denoted as t1 and t2, and the second target timeliness dependency rule contains eight dependency rules, denoted as r1…r8.

[0124]

[0125] Table 1

[0126]

[0127] Table 2

[0128] For example, Table 1 shows a process flow example provided by the present invention, and Table 2 shows a preset target timeliness dependency rule example. The process flow contains two tuples, denoted as t1 and t2, and each tuple contains multiple processes; the second target timeliness dependency rule contains eight dependency rules, denoted as r1…r8.

[0129] Based on the data in Table 1 and rules r1...r5, we can draw the conclusion t. i < High voltage test t j Directly dependent on t i < Load test t j , t < high voltage test 2023 directly depends on t < Load test 2023;

[0130] However, according to rules r1…r5, in actual industrial production, there are situations where the dependency strength between processes varies. In other words, some processes have strong, definite dependencies on each other's data, or between processes and product attributes, while others have lower dependency strength. Furthermore, because the dependencies or rules provided by users are highly subjective, two rules may simultaneously lead to the same conclusion but with different dependency strengths. To address this, we select the rule with the higher dependency strength as the final dependency strength of the conclusion. This is because the dependency strength of a rule is defined as the probability that the dependency is true; a higher probability often carries stronger persuasiveness. This can be expressed as: Where tml(rhs(r))=tml(r) i )×tml(lhs(r i For example, two dependency rules derive t respectively. i < High voltage test t j If the probabilities are 80% and 50%, then we ultimately give t. i < High voltage test t j The probability is 80%. The rationale for choosing the rule with higher dependency strength is that the left-hand side of the rule is equivalent to a condition, so we are more willing to believe the conclusion drawn from the dependency rule with higher probability.

[0131] Based on rules r6 and r7, we can arrive at the conclusion t. i < Load test t j and t i < Higg voltage test t j They all directly depend on t i < Vibration test t j Specific dependencies are as follows: Figure 3 As shown, Figure 3 This is a schematic diagram of a conclusion dependency sequence diagram provided in an embodiment of the present invention. From Figure 3 In this process, the dependencies between the conclusions of each process can be determined.

[0132] According to rule r8, t i < Load test t j With t j < Load test New conclusions were reached through cooperation in 2023. i < load test 2023, thus revealing the implicit dependencies. For example... Figure 4 The dashed box indicates that, based on rule r8, t i < Load test t j With t j < Load test New conclusions were reached through cooperation in 2023. i < Load test 2023.

[0133] like Figure 5 The diagram shown is a flowchart of a data quality assessment method provided in an embodiment of the present invention. (Refer to...) Figure 5 As shown, firstly, a conclusion dependency sequence graph is constructed from the acquired data. Then, based on the process corresponding to each conclusion in the conclusion dependency sequence graph, conclusion nodes with the same process are merged to obtain a process dependency sequence graph. After topological sorting of the process dependency sequence graph, the timeliness calculation order of each process data is obtained. Based on the timeliness calculation order of each process data, the throughput value between each process is calculated to determine the timeliness between each process. Finally, based on the timeliness between each process, the timeliness of each tuple and the dataset is obtained through weighted summation.

[0134] This invention proposes a manufacturing big data quality assessment method based on timeliness dependency rules. This method determines the non-deterministic dependencies between different processes based on the non-deterministic dependencies between conclusions of various data points. Then, based on target timeliness dependency rules and the non-deterministic dependencies between processes, a weighted timeliness graph is constructed. The timeliness of the current data at the target time is determined based on this weighted timeliness graph. The absolute values ​​of the differences between the current data's timeliness at the target time and a first value (a lower limit of the preset timeliness assessment value) and a second value (a higher limit of the preset timeliness assessment value) are compared. If the absolute value of the first difference is not greater than the absolute value of the second difference, the current data is deemed invalid. If the absolute value of the first difference is greater than the absolute value of the second difference, the current data is deemed valid. This manufacturing big data quality assessment method based on timeliness dependency rules overcomes the limitations of timestamps on data quality assessment. It can determine the timeliness between data points solely based on data timeliness dependency rules even in the absence of timestamps, and ultimately effectively assess the quality of manufacturing big data from the perspective of timeliness.

[0135] The various embodiments in this specification are described in a progressive manner, with each embodiment focusing on the differences from other embodiments. The same or similar parts between the various embodiments can be referred to each other.

[0136] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, apparatus, or computer program products. Therefore, embodiments of the present invention can take the form of entirely hardware embodiments, entirely software embodiments, or embodiments combining software and hardware aspects. Furthermore, embodiments of the present invention can take the form of computer program products implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0137] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, terminal devices (systems), and computer program products according to embodiments of the invention. It should be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a general-purpose computer, special-purpose computer, embedded processor, or other programmable processor of a manufacturing big data quality assessment terminal device based on time-bound rules to produce a machine, such that the instructions, which execute via the computer or other programmable processor of the manufacturing big data quality assessment terminal device based on time-bound rules, generate instructions for implementing the process... Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0138] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable manufacturing big data quality assessment terminal device based on time-cycle rules to operate in a specific manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in the process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0139] These computer program instructions can also be loaded onto a computer or other programmable manufacturing big data quality assessment terminal device based on time-bound cycle rules, causing a series of operational steps to be executed on the computer or other programmable terminal device to produce computer-implemented processing, thereby providing instructions that execute on the computer or other programmable terminal device for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0140] Although preferred embodiments of the present invention have been described, those skilled in the art, upon learning the basic inventive concept, can make other changes and modifications to these embodiments. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments as well as all changes and modifications falling within the scope of the embodiments of the present invention.

[0141] Finally, it should be noted that in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or terminal device that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or terminal device. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or terminal device that includes said element.

[0142] The above provides a detailed description of a manufacturing big data quality assessment method based on time-dependent rules provided by the present invention. Specific examples have been used to illustrate the principles and implementation methods of the present invention. The descriptions of the above embodiments are only for the purpose of helping to understand the method and core ideas of the present invention. At the same time, for those skilled in the art, there will be changes in the specific implementation methods and application scope based on the ideas of the present invention. Therefore, the content of this specification should not be construed as a limitation of the present invention.

Claims

1. A method for manufacturing big data quality authentication based on aging-dependent rules, characterized by, The method comprises: constructing a conclusion dependency sequence graph based on conclusions between each two of a plurality of data in a data set obtained from an industrial database and non-deterministic dependency relationships between the conclusions, wherein each of the conclusions is taken as a conclusion node of the conclusion dependency sequence graph, and a non-deterministic dependency relationship between each of the conclusion nodes is represented by a directed edge; merging conclusion nodes containing the same process in the conclusion dependency sequence graph to construct a process dependency sequence graph, wherein a process corresponding to a merged conclusion node is taken as a process node of the process dependency sequence graph, and a non-deterministic dependency relationship between each of the process nodes is represented by a directed edge; constructing a weighted timeliness graph of a target process based on a target timeliness dependency rule, taking a target time in the target timeliness dependency rule as an initial node, and taking each data in the target process as an intermediate node; determining timeliness of current data in the target process at the target time according to the weighted timeliness graph; comparing absolute values of a first difference between the timeliness of the current data at the target time and a preset timeliness appraisal lower limit and a second difference between the timeliness of the current data at the target time and a preset timeliness appraisal upper limit; in a case where the absolute value of the first difference is not greater than the absolute value of the second difference, identifying that the timeliness of the current data is invalid; in a case where the absolute value of the first difference is greater than the absolute value of the second difference, identifying that the timeliness of the current data is valid.

2. The time-dependent rule-based manufacturing big data quality appraisal method according to claim 1, wherein, The determination of the timeliness of the current data in the target process at the target time according to the weighted timeliness graph comprises: calculating the timeliness of the current data at the target time according to the following formula: wherein, is the timeliness of the current data at the target time; denotes data in the process preceding the target time of the conclusion; denotes data in the process whose timeliness is valid until the target time of the conclusion; is the current data; is the process; is the target time.

3. The time-dependent rule-based manufacturing big data quality appraisal method according to claim 2, wherein, A tuple comprises a plurality of processes, and the determination of the timeliness of the current tuple at the target time in the case of considering a tuple comprises: calculating the timeliness of the current tuple at the target time according to the following formula: wherein, is the current tuple, is the current tuple timeliness; denotes the current tuple different data each corresponding weight; denotes the timeliness of different processes.

4. The time-dependent rule-based manufacturing big data quality appraisal method according to claim 3, wherein, A data set comprises a plurality of tuples, and the determination of the timeliness of the current data set in the case of considering a data set comprises: calculating the timeliness of the current data set according to the following formula: wherein, is the current data set, is the current data set is the timeliness of the current data set; denotes the current data set different tuples each corresponding weight; denotes the timeliness of the different tuples.

5. The time-dependent rule-based manufacturing big data quality appraisal method according to claim 2, wherein, The determination of the dependency strength of a conclusion that the current data is valid before the target time comprises: calculating the dependency strength of a conclusion that the current data is valid before the target time in the target process according to the following formula: in, For target timeliness-dependent rules; Indicates the target time-dependent rule Order Indicate conclusion ; Indicate conclusion Target time-dependent rules The strength of the dependency that holds true; Pre-set timeliness rules for the target Order Indicates target time-dependent rules The conclusion on the left; Indicates target time-dependent rules The left-hand conclusion in the target time-dependent rule The strength of the dependency that holds true; Indicates target time-dependent rules Dependence strength; In a case where a plurality of target timeliness dependency rules exist, a maximum value of a plurality of dependency strengths of a conclusion that the current data is valid before the target time calculated based on the plurality of target timeliness dependency rules is determined as the dependency strength of the conclusion that the current data is valid before the target time.

6. The time-dependent rule-based manufacturing big data quality appraisal method according to claim 1, wherein, The determination of the timeliness of the current data in the target process at the target time according to the weighted timeliness graph comprises: According to the non-deterministic dependency relationship between the target time and the target data in the weighted aging graph, the flux value of the target time and the target data is determined as the aging property of the target data at the target time; the flux value represents the dependency strength of the non-deterministic dependency relationship between the target time and the target data; In the case of only one non-deterministic dependency relationship, the minimum flux value of the target time and the target data is determined as the aging property of the target data at the target time; or, In the case of at least two non-deterministic dependency relationships, the maximum value of the flux values of the target time and the target data under at least two non-deterministic dependency relationships is determined as the aging property between the target time and the target data.

7. The time-dependent rule-based manufacturing big data quality appraisal method according to claim 1, wherein, According to the target aging dependency rule, the target time in the target aging dependency rule is taken as an initial node, and each data under the target process is taken as an intermediate node to construct a weighted aging graph of the target process, specifically including the following steps: S1, obtaining an aging dependency rule set, and determining a first target aging dependency rule containing the target time from the aging dependency rule set; S2, determining a second target aging dependency rule based on the aging dependency rule set; S3, taking the target time in the first target aging dependency rule as an initial node, and taking each data in the target process that can satisfy the second target aging dependency rule as an intermediate node; S4, adding a directed edge between any two nodes with a non-deterministic dependency relationship in the initial node and the intermediate node to construct a weighted aging graph of the target process.

8. The time-dependent rule-based manufacturing big data quality appraisal method according to claim 1, wherein, Based on the weighted aging graph, the aging property of each data under the target process is determined, including: Based on the non-deterministic dependency relationship between any two data in each data under the target process, the respective conclusions of any two data are obtained; Based on the target aging dependency rule, the dependency strength of the respective conclusions of any two data is calculated; According to the dependency strength of the respective conclusions of any two data, the aging property of each data under the target process is determined. 9.The time-dependent rule based manufacturing big data quality appraisal method according to claim 8, wherein, The target aging dependency rule includes: a first dependency rule, a second dependency rule, a third dependency rule, a fourth dependency rule, a fifth dependency rule, a sixth dependency rule, and a seventh dependency rule; obtaining the conclusion between each two data in the plurality of data in the data set includes: Based on the first dependency rule, the second dependency rule, the third dependency rule, the fourth dependency rule, the fifth dependency rule, the sixth dependency rule, and the seventh dependency rule, the explicit non-deterministic dependency relationship between each two data in the plurality of data in the data set is determined; The explicit non-deterministic dependency relationship between each two data in the plurality of data in the data set is taken as the conclusion between each two data in the plurality of data in the data set. 10.The time-dependent rule based manufacturing big data quality appraisal method according to claim 7, wherein, The second target aging dependency rule further includes an eighth dependency rule; the method further includes: Based on the eighth dependency rule, determine an implied non-deterministic dependency relationship between each two of the plurality of data in the data set. Based on the eighth dependency rule, determine an implied non-deterministic dependency relationship between each two of the plurality of data in the data set. Based on the eighth dependency rule, determine an implied non-deterministic

Images