Data processing system, method and storage medium based on diagnostic task flow constraints

Abstract

The application provides a data processing system and method based on diagnostic task flow constraints and a storage medium, and belongs to the technical field of industrial Internet of Things and intelligent decision-making. The system comprises a storage medium and a processor. The processor executes a pipeline control program, a task execution program and a context management program. The pipeline control program controls node activation according to the logical completeness of the structured intermediate result output by the predecessor node and the argument reliability index through a gate check program. The gate check program generates the specific rules or thresholds as the verification basis information. The context management program stores the verification basis information, the corresponding intermediate result and the verification conclusion in the diagnostic context vector, so that they become data elements that can be called by subsequent nodes. The application converts the scheduling decision logic into auditable data, and solves the technical problem that the diagnostic process cannot be verified because the decision basis is not traceable.

Description

Technical Field

[0001] This invention relates to the fields of industrial Internet of Things and intelligent decision-making technology, specifically to a data processing system, method and storage medium for anomaly diagnosis and root cause analysis of complex systems, and particularly to an automated task flow execution technology that can ensure logical coherence and data traceability in the diagnostic process. Background Technology

[0002] In fields such as industrial production, energy management, and equipment operation and maintenance, rapid and accurate diagnosis of system anomalies or potential hazards is crucial for ensuring safety and efficiency. Currently, task orchestration tools or complex event processing systems with a certain degree of automation can manage the order and dependencies of tasks, but their core mechanisms have inherent technical limitations when applied to diagnostic scenarios.

[0003] The limitations of existing technology are mainly twofold:

[0004] First, the decision-making basis for task scheduling is disconnected from the intrinsic quality of diagnostic conclusions, and the decision-making process itself lacks logical transparency. The "gating" of general systems is typically based solely on task execution status codes or simple data format checks. However, in diagnostic scenarios, the output of an analysis step is itself a conclusion with probabilistic and multidimensional attributes. Existing technologies lack a mechanism to deeply evaluate the logical completeness and credibility of the conclusion itself, and to use this as a direct basis for subsequent scheduling. More importantly, even if some quality check is performed, the crucial logic of "which specific rule and threshold is used to make the pass / fail decision" is usually untraceable after the process ends. This results in the scheduling decision itself being unauditable, lacking a mechanism for maintaining intermediate states and decision-making basis.

[0005] Second, the storage model of process data is inconsistent with the need to construct an auditable chain of reasoning evidence, lacking a solidification of decision-making basis. Existing systems typically only record task inputs and outputs or final states, without recording the specific criteria supporting the validity of intermediate conclusions. The reliability of diagnostic conclusions depends on a complete and traceable chain of reasoning evidence, which should include not only "what" but also "why." Current technology lacks a mechanism to synchronously and structurally persist the specific evidence and conclusions of quality verification during the diagnostic process, thus failing to construct an independently verifiable evidence archive containing complete decision-making logic.

[0006] Therefore, when general workflow or task scheduling techniques are directly applied to the field of professional diagnosis, technical problems arise, such as the inability to trace the basis for scheduling decisions, which leads to the inability to audit the diagnosis process and the difficulty in verifying the credibility of the conclusions. Summary of the Invention

[0007] The technical problem to be solved by this invention is to overcome the technical defects of existing general workflow or task scheduling technologies when applied to professional diagnostic scenarios, which are such that the scheduling decision basis is not traceable, resulting in the process being unauditable and the credibility of the conclusions being difficult to verify.

[0008] To solve the above-mentioned technical problems, the present invention proposes the following technical solution:

[0009] A data processing system based on diagnostic task flow constraints is proposed. Its core lies in constructing a technical architecture through hardware and software collaboration that synchronously solidifies the scheduling decision-making process and its complete basis, thereby building an auditable chain of evidence. The system mainly includes storage media and a processor.

[0010] The storage medium is used to store a predefined directed task flow graph. This graph is a data structure that encodes a specific diagnostic methodology into multiple interdependent diagnostic task nodes.

[0011] The processor, serving as the system's computing and control core, is programmed to execute multiple logic programs:

[0012] 1. Pipeline Control Program: This program reads the directed task flow graph to determine the topological order between task nodes and executes a gating verification program as the core decision-making mechanism for scheduling. Before activating each subsequent node, the gating verification program reviews the structured intermediate results output by the predecessor node, verifying its logical completeness indicators and / or assertion credibility indicators. The improvement of this invention includes: the gating verification program not only outputs a "pass / fail" conclusion, but also explicitly generates the specific rules, thresholds, or model identifiers on which its verification is based as verification basis information. This allows the internal logical parameters of the scheduling decision to be transformed into processable data in real time.

[0013] 2. Task Execution Program: This program is responsible for scheduling specific analysis operators to execute activated diagnostic task nodes and generate new structured intermediate results.

[0014] 3. Context Management Program: This program allocates and maintains a dedicated diagnostic context vector in the system memory for each independent diagnostic session. In this invention, the "diagnostic context vector" refers to the core runtime state carrier created and maintained by the system throughout the entire lifecycle of a single diagnostic session, serving as the real-time data transfer and sharing carrier between various diagnostic task nodes. It is the core state container driving the diagnostic process forward, rather than a log archived afterward. Further improvements to this invention include: the program is specifically configured to associate and store the verification basis information generated by the gating verification program with the intermediate results that triggered the verification and the conclusion of the verification. This transforms the decision parameters of the control flow into part of the data flow and integrates them into the same carrier. Simultaneously, the vector also stores all intermediate results, node identifiers, and logical dependencies sequentially. Therefore, the diagnostic context vector not only records the data flow but also comprehensively records the precise basis of the control flow (scheduling decisions), enabling each path selection to be independently reproduced and verified afterward, and its stored basis information can be actively invoked by subsequent nodes (such as fusion analysis nodes), forming a closed loop of "decision-basis-re-decision".

[0015] The solution of this invention realizes the transformation of process control logic into auditable data assets. Verification basis information is generated through a gating verification procedure, and then synchronized and solidified into the core runtime carrier along with process data by a context management procedure. The system generates a complete evidence package for the diagnostic process, containing "data-conclusion-rules for drawing conclusions." This differs from general methods that only record post-event logs of states or only store isolated quality inspection results.

[0016] Furthermore, the processor can also execute a data fusion program to extract multi-source intermediate results and their related verification basis information from the diagnostic context vector when the fusion analysis task node is triggered, and perform cross-validation and weighted calculation.

[0017] Furthermore, the processor can also execute a feedback learning program, using historical diagnostic context vectors and processing feedback data to optimize and write back various parameters in the system, thereby achieving closed-loop evolution.

[0018] In addition, the processor can also execute a report generation program to generate a verification report that deeply reflects the process logic and decision-making basis based on the diagnostic context vector.

[0019] Based on the same inventive concept, the present invention also provides a corresponding diagnostic data processing device, a data processing method, and a computer-readable storage medium storing a computer program that implements the above-mentioned system functions. Beneficial effects

[0020] Compared with the prior art, the technical solution of the present invention can produce the following beneficial effects:

[0021] 1. Achieved deep traceability and auditability of the diagnostic scheduling decision-making process: By explicitly generating and solidifying "verification basis information" into the core runtime data carrier through a gating verification program, the system, for the first time, transforms the specific technical basis of scheduling decisions into queryable, verifiable, and subsequently callable data. This fundamentally changes the way process auditing is conducted, making it possible to examine "why this scheduling choice was made," effectively solving the problem of the invisibility of internal decision-making processes in diagnostic automation.

[0022] 2. A complete chain of diagnostic evidence containing causal relationships has been constructed: The context management program links and stores verification evidence, intermediate results, and verification conclusions in the core carrier, making the final diagnostic context vector a self-contained evidence unit with causal explanatory capabilities. It not only shows "what happened" but also explains "why it happened," providing unprecedented data support for liability determination, compliance verification, and conclusion credibility assessment.

[0023] 3. Improved reliability and credibility of complex diagnostic systems: Scheduling based on clear, traceable, and reusable decision criteria reduces the risk of erroneous conclusions in individual steps contaminating the entire reasoning chain. Simultaneously, it provides a high-quality, highly interpretable training data foundation for subsequent data fusion and feedback learning, contributing to continuous optimization of system performance.

[0024] 4. Promotes the evolution of automated diagnostic systems towards a highly reliable and accountable paradigm: This solution transforms the process control logic of software into persistently auditable data assets, providing key technical means for achieving quality control and trusted certification throughout the entire diagnostic lifecycle, which meets the stringent requirements of highly reliable industrial fields for transparent, trustworthy, and intelligent systems. Attached Figure Description

[0025] Figure 1 This is a schematic diagram of the hardware and logic module architecture of a system in one embodiment of the present invention.

[0026] Figure 2 This is an example diagram of the Directed Task Flow Graph (DAG) of this invention.

[0027] Figure 3 This is a schematic diagram of the data structure of the diagnostic context vector in this invention.

[0028] Figure 4 This is a schematic diagram of the gating verification program execution flow in this invention.

[0029] Figure 5 This is a schematic diagram of the data fusion program execution flow in this invention.

[0030] Figure 6This is a schematic diagram of the overall flow of the diagnostic data processing method in Embodiment 1 of the present invention.

[0031] Figure 7 This is a logical diagram of the confidence assessment rule in Embodiment 1 of the present invention.

[0032] Figure 8 This is a schematic diagram of the equipment fault diagnosis task flow in Embodiment 2 of the present invention. Detailed Implementation

[0033] To make the objectives, technical solutions, and advantages of the present invention clearer, the present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0034] Example 1: Closed-loop diagnosis of hidden dangers applied to industrial safety management

[0035] This embodiment uses the diagnosis of the potential danger of "failure of on-site fire-fighting equipment" in a thermal power plant as an example to illustrate the specific working method of this system.

[0036] 1. System initialization and triggering:

[0037] (1) The safety officer discovered that the pressure gauge pointer of the fire extinguisher in boiler room No. 3 was in the red zone through the explosion-proof handheld terminal. He took a photo using the terminal APP and described it verbally.

[0038] (2) The handheld terminal sends the diagnostic trigger command to the server.

[0039] (3) The server processor runs the pipeline control program and creates a session ID. The context management program initializes a diagnostic context vector, which serves as the core runtime state carrier for this session.

[0040] 2. Solidification of task flow execution and decision-making basis:

[0041] (1) Step S1 (Phenomenon Feature Extraction): The controller loads the task flow graph ( Figure 2 The starting node S1 is activated and executed, outputting a structured result, such as {"node_id": "S1", "device": "dry_powder_extinguisher", … ,"confidence": 0.96} to the diagnostic context vector.

[0042] (2) Step S2 (system compliance check): The controller determines that the S2 node needs to be activated. First, the gate control verification program is executed to verify the output of S1.

[0043] Logical completeness verification: The program checks according to the predefined rule "Required field list: ['device', 'location', 'status', 'confidence']". The specific rule used for this verification is generated as part of the verification basis information (denoted as rule_001).

[0044] Statement credibility verification: The program calls the confidence assessment program, which evaluates the statement according to the rule "IF confidence >= 0.85 THEN PASS". The content of this rule and the threshold of 0.85 are generated as another part of the verification basis information (denoted as rule_002).

[0045] The gate verification result is "passed".

[0046] (3) The context management program associates and stores the above verification basis information (rule_001, rule_002), the S1 output data snapshot that triggered the verification, and the verification conclusion "pass" in the verification_log field of the diagnostic context vector (see Figure 3 This information has become part of the diagnostic context vector and can be invoked by any subsequent node query.

[0047] (4) The S2 node is activated and executed, outputting its diagnostic results and updating the vector.

[0048] 3. Data Fusion and Root Cause Analysis (Supplement to Key Implementation Examples):

[0049] (1) The process proceeds to the fusion analysis task node S5. The data fusion program reads the intermediate results of S1, S2, and S4 from the diagnostic context vector.

[0050] (2) S5 Fusion Analysis Operator Execution Processing. As a key embodiment of the closed loop of the present invention, before performing weighted calculation, the operator first queries and reads the verification basis information (i.e., rule_001, rule_002 and their corresponding thresholds) of nodes S1, S2, and S4 from the diagnostic context vector. The internal logic of the operator is adjusted accordingly: when the verification basis information indicates that the conclusion of a certain predecessor node (such as S1) is based on a high-strictness threshold (such as 0.85 in rule_002) and is obtained through verification, a higher weight is assigned to the conclusion (confidence: 0.96) of that node in the fusion calculation; otherwise, its weight is reduced or additional verification rules are triggered. This process confirms that the "verification basis information" is actively "called" by subsequent nodes and directly affects the decision.

[0051] (3) The weighted calculation can be expressed as C_fused = Σ (w_i × C_i). The value of the weight w_i (e.g., W1 = 0.3) can be dynamically adjusted based on the above analysis of the verification basis information. The final confidence level is: 0.3 × 0.96 + 0.4 × 1.0 + 0.3 × 0.90 = 0.958.

[0052] (4) Output the fusion conclusion.

[0053] 4. Report generation and learning loop:

[0054] (1) The report generation program generates a report, which clearly shows the basis for the activation of the S2 node and the basis for adjusting the weights of each input when the S5 node is fused.

[0055] (2) The feedback learning program optimizes parameters based on historical data (such as adjusting the threshold of rule_002) and loads the new parameters to form a closed loop.

[0056] Example 2: Predictive maintenance diagnosis applied to wind power equipment

[0057] Draw the diagnostic task flow diagram through the graphical configuration interface. Figure 8 After the system is loaded, it runs. Its core process is similar to that of Example 1. The difference lies in the fact that the specific diagnostic rules, analysis operators and fusion logic are designed around the characteristics of equipment faults. The rules and basis information for its gating verification change accordingly, but the architecture mechanism for generating and solidifying the decision basis into the core state carrier at runtime and making it available for subsequent process calls remains the same.

[0058] Summary statement

[0059] The above embodiments are for illustrative purposes only and are not intended to limit the scope of the invention. Any substitutions, modifications, or variations based on the core concept of this invention—namely, explicitly generating specific basis information for verification decisions in diagnostic task flow scheduling and synchronously solidifying it with process data into the diagnostic context vector, which serves as the core state carrier between nodes during runtime, to construct an auditable and closed-loop chain of evidence—such as employing different rule representations, storage formats, or application areas, should be covered within the scope of protection of this invention.

Claims

1. A data processing system based on diagnostic task flow constraints, characterized in that, include: The storage medium stores a predefined directed task flow graph, which defines multiple diagnostic task nodes and their execution dependencies. The processor, communicatively connected to the storage medium, is configured to execute: A pipeline control program is used to control the activation order of the diagnostic task nodes according to the directed task flow graph and a gating verification program; wherein, the gating verification program is configured to: verify the structured intermediate results output by at least one predecessor diagnostic task node before activating the current diagnostic task node, the verification including at least the evaluation of the logical completeness indicators and / or confidence indicators of the conclusions contained in the structured intermediate results; and the gating verification program generates verification basis information as the specific rules, thresholds or model identifiers on which it is based. The task execution program is used to call the corresponding analysis operator to execute the activated diagnostic task node and generate new structured intermediate results; A context management program is used to create and maintain a diagnostic context vector in the system memory for a single diagnostic session. The diagnostic context vector constitutes the core runtime state carrier for real-time data transmission and sharing among diagnostic task nodes during a diagnostic session. The context management program is configured to associate and store the verification basis information generated by the gating verification program, the structured intermediate results that triggered the verification, and the conclusion of the verification in the diagnostic context vector, so that the verification basis information becomes a data element in the diagnostic context vector that can be called by subsequent diagnostic task nodes. The diagnostic context vector also stores all generated structured intermediate results, the identifiers of the diagnostic task nodes that generated each result, and the logical dependencies between the results in sequence.

2. The system according to claim 1, characterized in that, The logical completeness index is determined by verifying whether the structured intermediate result contains predefined required data fields; the assertion credibility index is calculated by a confidence assessment program based on the content quality of the structured intermediate result or directly provided by the analysis operator that generates the result.

3. The system according to claim 1, characterized in that, The processor is also configured to execute a data fusion program; The directed task flow graph defines at least one fusion analysis task node; When the pipeline control program activates the fusion analysis task node, the data fusion program is triggered to read at least two structured intermediate results generated by different historical diagnostic task nodes from the diagnostic context vector, and input the read intermediate results into the fusion analysis operator corresponding to the fusion analysis task node for processing.

4. The system according to claim 3, characterized in that, The data fusion program is configured to determine the historical diagnostic task nodes to be read based on the logical dependencies stored in the diagnostic context vector.

5. The system according to claim 3, characterized in that, The data fusion program includes a weight calculation subroutine, which dynamically adjusts the contribution weight of each of the multiple intermediate results read in the fusion analysis operator processing process based on the source node type and / or associated assertion credibility index of each intermediate result.

6. The system according to claim 1, characterized in that, The processor is also configured to execute a feedback learning program; The feedback learning procedure is configured as follows: Obtain feedback data on the handling measures corresponding to the completed diagnostic sessions; The feedback data of the treatment measures are associated and stored with the corresponding diagnostic context vector to form a training sample set; Using the training sample set, the parameters of the analysis operator model in the task execution program, the contribution weight in the data fusion program, or the evaluation threshold in the gating verification program are optimized, and the optimized parameters are loaded into the corresponding programs.

7. The system according to claim 1, characterized in that, The predefined directed task flow graph is generated by receiving drag-and-drop, connection, and parameter setting operations from the user through a graphical configuration interface.

8. The system according to claim 1, characterized in that, The processor is also configured to execute a report generation program to generate a verification report reflecting the logical integrity of the diagnostic process and the traceability of the decision basis, based on the intermediate result sequence, logical dependencies and verification basis information recorded in the diagnostic context vector.

9. A diagnostic data processing device, characterized in that, include: Memory is used to store predefined directed task flow graphs and computer programs; A processor, coupled to the memory, is used to execute the computer program to perform the functions of all programs executed by the processor in the system as claimed in any one of claims 1 to 8.

10. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is executed by the processor, it performs the functions of all programs executed by the processor in the system as described in any one of claims 1 to 8.

11. A data processing method based on diagnostic task flow constraints, characterized in that, The method operates on the system as described in any one of claims 1 to 8, and includes the following steps: Step S10: In response to the diagnostic trigger command, a diagnostic session is created, and the corresponding diagnostic context vector is initialized in the system memory. The diagnostic context vector constitutes the core runtime state carrier for real-time data transmission and sharing among the diagnostic task nodes in this session. Step S20: Based on the predefined directed task flow graph, determine the diagnostic task node to be executed and perform gating verification; the gating verification includes: verifying whether the outputs of all predecessor nodes of the node have been stored as structured intermediate results in the diagnostic context vector, and verifying the logical completeness index and / or assertion credibility index contained in the output; the gating verification generates verification basis information by using the specific rules, thresholds or model identifiers on which it is based. Step S30: If the gating verification passes, the analysis operator corresponding to the node is called to execute the diagnostic logic and generate the structured intermediate result of the current node; Step S40: Write the structured intermediate result of the current node and its associated diagnostic task node identifier into the diagnostic context vector, and update the logical dependency relationship; at the same time, store the verification basis information, the intermediate result that triggered the verification, and the verification conclusion generated in step S20 in the diagnostic context vector, so that they become data elements that can be called by subsequent diagnostic task nodes. Step S50: Repeat steps S20 to S40 until all nodes in the directed task flow graph have been executed.

12. The method according to claim 11, characterized in that, It also includes step S60: generating and outputting a logical integrity verification report based on the final complete diagnostic context vector.

13. The method according to claim 11, characterized in that, In step S20, if the node to be executed is a fusion analysis task node, then step S30 includes: Sub-step S31: Read at least two structured intermediate results generated by different historical diagnostic task nodes from the diagnostic context vector; Sub-step S32: Take the multiple intermediate results read as joint input and call the fusion analysis operator for processing; wherein, the fusion analysis operator performs logical reasoning based on a predefined production rule knowledge base, and calculates the weighted confidence level based on the assertion credibility index associated with each intermediate result to generate the fusion analysis result.

Images