A power battery transportation intelligent compliance method and system

Abstract

This invention discloses an intelligent compliance method and system for power battery transportation, relating to the field of logistics management technology. The method includes: acquiring transportation request information containing battery parameters, packaging parameters, status parameters, supporting document parameters, transportation mode parameters, destination parameters, and applicable regulatory time parameters; constructing a regulatory context, retrieving structured regulatory facts from a regulatory knowledge graph within this context, and determining a set of candidate compliance rules after adjudication based on rule priority and coverage information; standardizing the transportation request information and structured regulatory facts into fact objects and performing forward chain reasoning, outputting compliance judgment results, a set of compliance actions, compliance solutions, and a regulatory basis tracing path. Through this invention's technical solution, automatic judgment and solution output across transportation modes are achieved, reducing regulatory mismatches and omissions, improving consistency, interpretability, and auditability, and enhancing operational efficiency.

Description

Technical Field

[0001] This invention relates to the field of logistics management technology, and in particular to an intelligent compliance method and system for transporting power batteries. Background Technology

[0002] Current compliance efforts for the cross-border transportation of power batteries (such as lithium batteries) heavily rely on manual comparison and experience-based judgment of the UN's Recommendations on the Transport of Dangerous Goods and its derivative regulations (such as IMDG, IATA, and ADR). While mainstream international logistics / packing software covers basic dangerous goods classifications, it typically requires users to manually configure templates in the lithium battery field. It lacks a built-in logic engine for UN38.3 testing prerequisites, UN number and packaging description mapping, and clause-level requirements, resulting in compliance determinations still primarily being done manually, and parameter entry and rule selection are prone to errors.

[0003] Meanwhile, UN regulations exhibit coupled constraints across the "testing-packaging-transportation mode" dimension: the watt-hour thresholds, packaging segmentation clauses, labeling, and documentation requirements differ for the same UN number in air transport (IATA), ocean transport (IMDG), and land transport (ADR) scenarios, causing compliance solutions to change with the mode of transport. Traditional keyword-based regulatory queries or static rule tables struggle to express and handle these dynamic dependencies, easily leading to missed detections or mismatches, especially when dealing with strong premise-consequence dependency chains between UN3481 and UN38.3.

[0004] Furthermore, existing compliance outputs mostly remain at the level of conclusions or lists, lacking a mechanism to establish a traceable link between triggered rule identifiers and specific regulatory clause numbers. This makes it difficult to meet the requirements of internal audits, customer reviews, and regulatory spot checks for an explainable and reproducible chain of compliance evidence. Once regulatory versions are updated or business parameters change, it is also difficult to quickly locate the source of the differences between the affected rules and outputs, leading to increased compliance management costs. Summary of the Invention

[0005] To address the aforementioned issues, this invention provides an intelligent compliance method and system for power battery transportation. It determines the applicable set of regulations based on the transportation mode, retrieves structured regulatory facts associated with transportation request information from a regulatory knowledge graph, and then standardizes the transportation request information and structured regulatory facts into fact objects that a rule engine can process, performing forward chain reasoning. This enables the automatic output of compliance judgment results and corresponding packaging, labeling, and documentation requirements within a unified chain. By introducing clause identifiers into the structured regulatory facts and recording triggered rule identifiers during the reasoning process, a regulatory basis traceability path is generated, ensuring that compliance conclusions are interpretable, auditable, and verifiable. This reduces the risk of compliance disputes and improves the efficiency and consistency of compliance processing across transportation modes.

[0006] To achieve the above objectives, the present invention provides a smart compliance method for the transportation of power batteries, comprising: Obtain the transportation request information of the power battery to be transported. The transportation request information includes battery parameters, packaging parameters, status parameters, certification document parameters, transportation mode parameters, destination parameters, and applicable regulatory time parameters for determining the compliance of power battery transportation. A regulatory context is constructed based on the transportation mode parameters, the destination parameters, and the applicable time parameters of the regulations. This regulatory context is then injected as a context-binding variable into a context-aware graph query of the regulatory knowledge graph. Under this regulatory context, structured regulatory facts associated with the transportation request information are retrieved from the regulatory knowledge graph. The context-aware graph query includes at least dynamic filtering based on the destination's applicable scope and effective time, attribute value range queries based on battery parameters and packaging parameters, and meta-relational reasoning based on rule coverage relationships to obtain candidate regulatory facts and determine the coverage relationships in the candidate compliance rule set. Each compliance rule in the candidate compliance rule set carries rule priority information and coverage relationship information. The structured regulatory facts include clause identifiers, source document identifiers, and effective time information corresponding to the corresponding regulatory clauses. The transportation request information and the structured regulatory facts are standardized into fact objects supported by the rule engine. The fact objects are written into the rule engine and forward chain reasoning is performed based on the candidate compliance rule set after adjudication by rule priority information and coverage relationship information. The compliance judgment result is output and a set of compliance actions corresponding to the compliance judgment result is generated. Based on the compliance determination results and the set of compliance actions, compliance outputs are generated. The compliance outputs include compliance determination results and compliance solutions, wherein the compliance solutions include packaging requirements, labeling requirements, and documentation requirements. Output the compliance output and generate a legal basis tracing path corresponding to the compliance output. The legal basis tracing path includes the triggered rule identifier and the clause identifier and source document identifier associated with the rule identifier.

[0007] In the above technical solution, preferably, the structured regulatory facts are stored in the regulatory knowledge graph in the form of triples, and each triple carries a source document identifier attribute, a clause number attribute, and an effective time attribute. The source document identifier attribute, the clause number attribute, and the effective time attribute are used to establish a traceable association between the triggered rule identifier and the corresponding regulatory clause number, source document identifier, and regulatory effective time information when generating the regulatory basis tracing path. The structured regulatory facts include the correspondence between UN numbers and testing requirements, the mapping between UN numbers and packaging instruction entries, threshold constraint facts and applicable condition constraint facts divided into segmented thresholds and applicable conditions for packaging instruction entries, as well as geographical constraint facts corresponding to the scope of application of the destination and temporal constraint facts corresponding to the period of regulatory effectiveness.

[0008] In the above technical solution, preferably, the candidate compliance rule set consists of multiple compliance rules, each compliance rule carrying a mode of transport applicability attribute, a region applicability attribute, an effective time attribute, rule priority information, and coverage relationship information. The mode of transport applicability attribute is used to indicate the mode of transport to which the compliance rule applies, the region applicability attribute is used to indicate the destination area to which the compliance rule applies, and the effective time attribute is used to indicate the applicable time range of the compliance rule. The specific process of constructing a regulatory context based on the transportation mode parameters and obtaining the candidate compliance rule set includes: Based on the transport mode parameters, the regulatory context corresponding to the transport mode is determined, and combined with the destination parameters and the applicable time parameters of the regulations, a regulatory context including the transport mode, destination and applicable time of the regulations is formed; The regulatory context is injected as a context-binding variable into the context-aware graph query of the regulatory knowledge graph to retrieve structured regulatory facts that match the regulatory context. Based on the applicable attributes of the mode of transport, the applicable attributes of the region, and the effective time attributes, the compliance rules are dynamically filtered to obtain a set of candidate compliance rules that match the regulatory context. The candidate compliance rule set is adjudicated based on rule priority information and coverage relationship information to obtain a compliance rule subset, and the compliance rule subset is loaded into the rule engine for the forward chain reasoning.

[0009] In the above technical solution, preferably, the forward chain reasoning based on the candidate compliance rule set after adjudication via rule priority information and the coverage relationship information includes: performing reasoning in the order of preset rule groups, wherein the rule groups include at least a qualification review rule group, a packaging instruction selection rule group, and a label generation and document requirements rule group; The qualification review rule group is used to output the transportability conclusion in the compliance judgment result based on the test pass status and prohibition status. The packaging description selection rule group is used to output the packaging description item based on the battery parameters and packaging parameters in the transportation request information and the packaging description mapping relationship and applicable condition constraint facts in the structured regulatory facts. The label generation and document requirement rule group is used to output the label requirements and document requirements based on the compliance judgment result, the packaging description item, and the clause constraint information in the structured regulatory facts.

[0010] In the above technical solution, preferably, the compliance solution includes the UN number, packaging instructions, label list and documentation list; The regulatory tracing path includes a triggered rule identifier sequence, a snapshot of the fact object corresponding to the rule identifier sequence, and the regulatory clause number, source document identifier, and regulatory effective time information corresponding to the rule identifier sequence. The snapshot of the fact object includes at least the transportation mode parameters, destination parameters, and regulatory applicable time parameters involved in this compliance determination. Based on the rule identifier sequence, the regulatory clause number, the source document identifier, and the regulatory effective time information, a tracing explanation text is generated to output a human-readable tracing explanation. The tracing explanation text is used to characterize the chain of evidence for obtaining the compliance determination result in the context of the regulation.

[0011] The present invention also proposes an intelligent compliance system for power battery transportation, used to implement the intelligent compliance method for power battery transportation as described in any one of claims 1 to 5, comprising: a request information acquisition module, a regulatory context determination module, a regulatory knowledge graph retrieval module, a rule set filtering module, a fact standardization module, a rule engine reasoning module, a compliance output generation module, and a traceability generation module; The request information acquisition module is used to acquire the transportation request information of the power battery to be transported. The transportation request information includes battery parameters, packaging parameters, status parameters, certification document parameters, transportation mode parameters, destination parameters, and regulatory application time parameters for determining the compliance of power battery transportation. The regulatory context determination module is used to determine the regulatory context based on the transportation mode parameters, destination parameters, and regulatory applicable time parameters. The regulatory knowledge graph retrieval module is used to inject the regulatory context as a context binding variable into the context-aware graph query of the regulatory knowledge graph, and retrieve structured regulatory facts associated with the transport request information from the regulatory knowledge graph under the regulatory context. The structured regulatory facts include clause identifiers, source document identifiers and effective time information corresponding to the corresponding regulatory clauses. The rule set filtering module is used to dynamically filter compliance rules according to the regulatory context to obtain a candidate compliance rule set, and to adjudicate the candidate compliance rule set according to rule priority information and coverage relationship information to obtain an adjudicated candidate compliance rule set, and to load the adjudicated candidate compliance rule set into the rule engine inference module. The fact standardization module is used to standardize the transportation request information and the structured regulatory facts into fact objects supported by the rule engine and write them into the rule engine. The rule engine reasoning module is used to perform forward chain reasoning based on the adjudicated candidate compliance rule set, output compliance judgment results, and generate a set of compliance actions corresponding to the compliance judgment results; The compliance output generation module is used to generate compliance outputs based on the compliance judgment result and the set of compliance actions. The compliance outputs include the compliance judgment result and the compliance plan. The compliance plan includes packaging requirements, labeling requirements and documentation requirements. The traceability generation module is used to generate a legal basis traceability path corresponding to the compliance output. The legal basis traceability path includes the triggered rule identifier, the clause identifier associated with the rule identifier, the source document identifier, and the legal effective time information.

[0012] In the above technical solution, preferably, the regulatory knowledge graph retrieval module includes a triplet storage submodule and a clause attribute maintenance submodule; The triplet storage submodule is used to store the structured regulatory facts in the form of triples in the regulatory knowledge graph; The clause attribute maintenance submodule is used to configure source document identifier attribute, clause number attribute and effective time attribute for each triple, so as to establish a traceable association between the triggered rule identifier and the corresponding legal clause number, source document identifier and legal effective time information when generating the legal basis tracing path.

[0013] In the above technical solution, preferably, the rule set filtering module includes a rule attribute maintenance submodule and a rule filtering submodule; The rule attribute maintenance submodule is used to configure transportation mode applicable attributes, geographical applicable attributes, effective time attributes, and rule priority information for each compliant rule, and is also used to maintain the coverage relationship information between compliant rules; The rule filtering submodule is used to dynamically filter compliance rules carrying the applicable attributes of the mode of transport, the applicable attributes of the region, and the effective time attribute based on the regulatory context, to obtain a set of candidate compliance rules that match the regulatory context, and to adjudicate the set of candidate compliance rules based on the rule priority information and the coverage relationship information, to obtain a set of adjudicated candidate compliance rules, and to provide the set of adjudicated candidate compliance rules to the rule engine inference module.

[0014] In the above technical solution, preferably, the rule engine reasoning module executes forward chain reasoning in the order of preset rule groups, and executes the forward chain reasoning based on the candidate compliance rule set after being adjudicated by the candidate rule set screening and adjudication module after being adjudicated by rule priority information and coverage relationship information. The rule group includes at least the qualification review rule group, the packaging instruction selection rule group, and the label generation and document requirement rule group. The qualification review rule set is used to output transportability conclusions in the compliance judgment results based on the test pass status and prohibition status. The packaging description selection rule group is used to output packaging description entries based on the battery parameters and packaging parameters in the transportation request information, as well as the packaging description mapping relationship and applicable condition constraints in the structured regulatory facts. The label generation and documentation requirements rules group is used to output label and documentation requirements based on the compliance determination results, the packaging instructions, and the clause constraints in the structured regulatory facts.

[0015] In the above technical solution, preferably, the traceability generation module is used to generate a rule identifier sequence, a snapshot of the fact object corresponding to the rule identifier sequence, and information on the regulatory clause number, source document identifier, and regulatory effective time corresponding to the rule identifier sequence. The snapshot of the fact object includes at least the transportation mode parameter, destination parameter, and regulatory applicable time parameter involved in this compliance determination. Based on the rule identifier sequence, the regulatory clause number, the source document identifier, and the regulatory effective time information, a traceability description text is generated to output a human-readable traceability description. The traceability description text is used to characterize the chain of evidence for obtaining the compliance determination result in the context of the regulation.

[0016] Compared with the prior art, the beneficial effects of the present invention are as follows: (1) By determining the set of applicable regulations based on the transportation mode parameters, and retrieving the structured regulatory facts associated with the transportation request information from the regulatory knowledge graph under the set of applicable regulations, the compliance points scattered in different regulatory texts are aggregated in a structured manner, which supports the rapid matching of compliance rules in different scenarios such as air transport, sea transport, and land transport, and reduces the cost of manual retrieval and the risk of mismatch.

[0017] (2) By standardizing transportation request information and structured regulatory facts into fact objects supported by the rule engine, and loading a subset of compliance rules corresponding to the set of applicable regulations to perform forward chain reasoning, the compliance judgment result can be automatically output and a set of compliance actions corresponding to the judgment result can be generated, ensuring the logical closure between the compliance judgment and the subsequent handling plan, and reducing the business breakpoints where the judgment has a conclusion but no implementation measures.

[0018] (3) By generating compliance outputs based on compliance judgment results and compliance action sets, compliance solutions are structured and output in the form of packaging requirements, labeling requirements and document requirements. This allows for direct connection with pre-transportation preparation and operation execution, improving the feasibility and consistency of the solution and reducing configuration differences in specification switching and multi-batch processing.

[0019] (4) By generating a legal basis traceability path containing rule identifiers and clause identifiers, and further organizing rule trigger sequences and fact object snapshots to form traceability descriptions, the compliance conclusions are made traceable, interpretable and verifiable, which facilitates internal audit spot checks, customer delivery and regulatory verification, and reduces the operational risks caused by compliance disputes and unclear responsibilities. Attached Figure Description

[0020] Figure 1 This is a flowchart illustrating an intelligent compliance method for transporting power batteries according to an embodiment of the present invention. Detailed Implementation

[0021] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0022] The present invention will now be described in further detail with reference to the accompanying drawings: like Figure 1 As shown, a smart compliance method for power battery transportation provided by the present invention includes: Obtain the transportation request information of the power batteries to be transported, and organize the transportation request information into a field-based structure, including battery parameters, packaging parameters, status parameters, certification document parameters, transportation mode parameters, destination parameters, and applicable regulatory time parameters, so as to unify the input criteria for subsequent retrieval and inference.

[0023] A regulatory context is constructed based on transportation mode parameters, destination parameters, and applicable regulatory time parameters. This regulatory context is then injected as a context-binding variable into a context-aware graph query of the regulatory knowledge graph. Within this regulatory context, structured regulatory facts associated with the transportation request information are retrieved from the regulatory knowledge graph. The context-aware graph query includes at least: dynamic filtering based on destination applicability and effective time to extract regulatory nodes valid within the current destination and applicable time range; attribute value range queries based on battery and packaging parameters to perform inequalities or interval comparisons on numerical attributes (such as watt-hour thresholds and state of charge limits) in the regulatory knowledge graph, ensuring precise matching between candidate regulatory facts and current battery parameters; and meta-relational reasoning based on rule coverage relationships to identify and process coverage semantic relationships between regulatory nodes, automatically excluding candidate facts covered by higher-priority rules. Through these graph queries, candidate regulatory facts are obtained, and coverage relationships within the candidate compliance rule set are determined. Each compliance rule in the candidate compliance rule set carries rule priority information and coverage relationship information. Structured regulatory facts include clause identifiers, source document identifiers, and effective date information corresponding to the relevant regulatory clauses, ensuring a traceable link between the reasoning process and the regulatory basis at the clause granularity.

[0024] In an optional implementation, to further mathematically describe the matching process between the physical characteristics of the power battery in the transportation request information and the nodes of the regulatory knowledge graph, a formula for calculating the matching degree of the graph nodes based on the physical characteristics of the power battery can be used: Where M(n,f) represents the matching function between node n in the regulatory knowledge graph and the current power battery fact f, and ∧ represents the logical AND operator. f k Indicates the current power battery's first k The physical parameters include at least the rated watt-hours f. Wh Number of batteries in the package f Count and lithium metal content f LiContent , This represents the closed interval threshold defined in the regulatory node n. The indicative function takes the value 1 when the corresponding physical parameter falls within the closed interval threshold defined by the regulatory node, and 0 otherwise. This formula ensures that regulatory node matching is successful only when the physical characteristics of the power battery, such as its watt-hour count and quantity, simultaneously satisfy all threshold constraints defined by the regulatory knowledge graph node. This transforms attribute value range queries into a calculable technical means.

[0025] The transportation request information and structured regulatory facts are standardized into fact objects supported by the rule engine. The fact objects are written into the rule engine, and forward chain reasoning is performed based on the candidate compliance rule set after adjudication through rule priority information and coverage relationship information. The compliance judgment result is output, and a set of compliance actions corresponding to the compliance judgment result is generated, so that the judgment conclusion and the handling action form a closed loop.

[0026] Based on the compliance determination results and the set of compliance actions, compliance outputs are generated. The compliance outputs include compliance determination results and compliance solutions. The compliance solutions include packaging requirements, labeling requirements, and documentation requirements, thereby directly translating the reasoning results into actionable operational elements.

[0027] Finally, the system outputs compliance outputs and generates a legal basis traceability path corresponding to the compliance outputs. The legal basis traceability path includes the triggering rule identifier, as well as the clause identifier associated with the rule identifier, the source document identifier, and the legal effective date information.

[0028] In an optional implementation, to further quantify the compliance determination results, the system can also calculate the compliance determination confidence level based on the power battery attributes when outputting the compliance determination results. The compliance determination confidence level can be expressed as: in, Indicates the deviation of the physical parameters of the power battery. C doc This indicates that the file is missing a penalty item.

[0029] Among them, f Wh Θ represents the actual watt-hours of the power battery. limit This represents the upper limit of watt-hours stipulated in the currently applicable regulations. When the actual watt-hours do not exceed the upper limit of the regulations, the deviation of the physical parameter is 0 or close to 0, corresponding to a higher level of compliance confidence. When the actual watt-hours exceed the upper limit of the regulations, the deviation of the physical parameter increases, and the compliance confidence decreases exponentially. C doc Used to characterize the penalty for missing key supporting documents, when key supporting documents such as the UN38.3 test report are already available. C doc Set to 0 when key supporting documents are missing. C doc The maximum value is taken, causing the confidence level of the compliance judgment to drop rapidly to 0, thereby triggering a non-compliance judgment or manual review.

[0030] Specifically, the rule engine layer is used to store conditional-conclusion compliance rules generated based on factual data. These rules explicitly encode the following UN regulatory-specific logic: (1) Passing the UN38.3 test is a prerequisite for using UN3481; (2) Whether the battery is installed in the equipment determines whether PI966 or PI967 is applicable; (3) When the mode of transport is air, the "Cargo Aircraft Only" label must be added; (4) Even if the same UN number and the same battery parameters are applied, different priority compliance rules are allowed to be triggered if the destination region or the period of effective regulation is different. The rule conclusions that are valid in the current regulatory context are retained through the coverage relationship adjudication.

[0031] In this implementation, the omissions and mismatches caused by manual rule checking and matching are reduced by using regulatory context-driven graph retrieval, candidate rule screening and adjudication, rule engine forward reasoning, action set implementation, and clause-level and source-level tracing, and the auditability and verifiability of compliance conclusions are improved.

[0032] In the above implementation, preferably, the structured regulatory facts are stored in the regulatory knowledge graph in the form of triples, and each triple carries a source document identifier attribute, a clause number attribute, and an effective date attribute. When generating a regulatory basis tracing path, the source document identifier attribute, clause number attribute, and effective date attribute are used to establish a traceable association between the triggered rule identifier and the corresponding regulatory clause number, source document identifier, and regulatory effective date information, so that the tracing output not only locates the rule layer, but also the specific clause number, regulatory source, and regulatory effective period.

[0033] Structured regulatory facts include the correspondence between UN numbers and testing requirements, the mapping between UN numbers and packaging instructions entries, threshold constraints and applicable condition constraints of packaging instructions entries divided by segment thresholds and applicable conditions, as well as territorial constraints corresponding to the scope of application of the destination and temporal constraints corresponding to the period of effectiveness of the regulations.

[0034] Structured regulatory facts include: the correspondence between UN numbers and test requirements, used to express the constraint of "number-test preconditions" at the factual level, such as the correspondence between the UN 38.3 test standard and its eight subtests T.1-T.8; the mapping relationship between UN numbers and packaging instruction entries, used to express the mapping of "number-packaging instruction selection entry" at the factual level, such as UN 3481 corresponding to two packaging instructions, PI966 and PI967; and threshold constraint facts and applicable condition constraint facts of packaging instruction entries divided by segment thresholds and applicable conditions, used to express the branch constraints of the same packaging instruction entry in different threshold ranges and under different applicable conditions at the factual level, such as by Section The watt-hour threshold and applicable conditions for the I / II division; the geographical constraint facts corresponding to the scope of application of the destination, used to express the applicable geographical area of ​​regulations, supplementary provisions or regional exception rules, such as the requirement that specific German additional labels only apply to transportation scenarios with destinations in Germany or the EU; and the time constraint facts corresponding to the effective period of the regulations, used to mark the effective start and end time of the regulations at the fact layer, so that graph queries can accurately filter candidate regulations based on the applicable time parameter, and to express the application boundary of the same clause before and after the change of regulations.

[0035] In this implementation, the source document identifier attribute, clause number attribute, and effective time attribute are carried by the triple, which enables the tracing path to have clause-level positioning capability; the structured expression of threshold constraints and applicable condition constraints enables rule reasoning to maintain matching accuracy and consistency in multi-segment clause scenarios.

[0036] In the above implementation, preferably, the candidate compliance rule set consists of multiple compliance rules. Each compliance rule carries a transportation mode applicability attribute, a geographical applicability attribute, an effective time attribute, rule priority information, and coverage relationship information. The transportation mode applicability attribute indicates the transportation mode to which the compliance rule applies, the geographical applicability attribute indicates the destination area to which the compliance rule applies, and the effective time attribute indicates the applicable time range of the compliance rule. The rule priority information is used to determine the order of effectiveness when multiple rules conflict in the same scenario, and the coverage relationship information is used to explicitly declare the effect of a rule replacing another rule under specific conditions.

[0037] The specific process of constructing a regulatory context and obtaining a set of candidate compliance rules based on transportation mode parameters includes: Based on the mode of transport parameter, the regulatory context corresponding to the mode of transport is determined. Combined with the destination parameter and the applicable time parameter of the regulation, a regulatory context including mode of transport, destination and applicable time of regulation is formed, so that the scope of regulation and mode of transport, destination and version time parameters form a definite correspondence. The regulatory context is injected as a context-binding variable into the context-aware graph query of the regulatory knowledge graph to retrieve structured regulatory facts that match the regulatory context. Based on the applicable attributes of mode of transport, applicable regions, and effective time, the compliance rules are dynamically filtered to obtain a set of candidate compliance rules that match the regulatory context. Finally, the candidate compliance rule set is adjudicated based on rule priority information and coverage relationship information to resolve conflicts between candidate rules, obtain a compliance rule subset, and load the compliance rule subset into the rule engine for forward chain reasoning.

[0038] In one alternative implementation, to describe the process of performing forward chain reasoning on the adjudicated set of candidate compliance rules as a path search process on a regulatory knowledge graph, a compliance path potential accumulation formula based on graph traversal can be used: in, ε path (P) represents the total potential energy of a complete compliance path P, and m represents the number of regulatory nodes included in the compliance path. t represents the region-specific weight of the i-th regulatory node. now Indicates the current time. This indicates the effective time of the i-th regulatory node. M ( n i , f ) represents the current power battery fact f and the i-th regulatory node n. i The matching function between them.

[0039] Using this formula, the system can calculate the path potential energy for each of the possible candidate compliance paths and select the path with the highest potential energy as the rule link corresponding to the final compliance solution; when a regulatory node in a path does not match the current reality of the power battery, the corresponding... M ( n i , f The value of ) = 0, thus naturally inhibiting or interrupting the path during the accumulation of path potential energy. Furthermore, due to the region-specific weights... By assigning higher weights to regional rules or national-specific rules, the system can automatically select rule paths that are more specific in the region, closer in time to the current time, and more consistent with the current power battery situation during graph traversal.

[0040] During implementation, the same UN number (e.g., an 80Wh built-in lithium battery) will automatically load different candidate rules and complete the adjudication under different modes of transport, different destinations, and different applicable regulations. The compliance solutions may be completely different. The specific operating rules are shown in the table below: For example, when the regulatory context is: destination = Germany, mode of transport = air, and applicable date of the regulation is 2026-03-07, the context-aware graph query not only determines whether "air" matches the mode of transport, but also whether the applicable region of the relevant rule covers Germany or its region, whether the rule's effective date is earlier than or equal to 2026-03-07, and further determines whether there is a high-priority rule that covers the current candidate rule. If a high-priority rule that satisfies the current regulatory context exists, the covered rule will not enter the rule engine in this reasoning, thus preventing low-priority old rules from continuing to participate in the decision-making.

[0041] During implementation, context awareness employs context-bound clauses (WITH CONTEXT) to inject transport mode, destination, and applicable regulatory time into the graph query logic. A priority-aware matching mechanism sorts the results according to rule priority. Simultaneously, conflict resolution projection (RESOLVE BY) is performed based on coverage information. While returning a set of candidate compliance rules, the effective order of each rule is determined for UN rule applicability determination. Transport mode, destination, and applicable regulatory time serve as the meta-conditions for rule activation. The same UN number loads different rule sets under different transport modes, regions, and time points. Each UN rule, in addition to the applyToTransportMode attribute, may also include applyToJurisdiction and effectiveDate attributes. During inference, rules are filtered and coverage decisions are made based on the user-specified transport mode parameter according to the current regulatory context. The region applicability attribute filters rules applicable only to specific geographical areas based on the destination parameter, while the effective time attribute excludes rules that are not yet effective or have expired based on the applicable regulatory time parameter, ensuring that the system's output compliance solution accurately adapts to the current transport scenario.

[0042] For example, in a certain implementation scenario, if maritime regulations allow 80Wh built-in batteries to be subject to PI967 Section II, while a supplemental air transport rule implemented in a German region after a specified effective date imposes stricter requirements on similar batteries, then when the regulatory context meets the requirements of Germany, air transport, and the corresponding effective date, the system will prioritize activating the rules from the region with higher priority, and will mask the baseline rules through coverage information. Conversely, if the regulatory context does not meet the requirements for that region or the effective date has not yet been reached, the system will retain the baseline rules. Therefore, the retrieval constraints formed by the regulatory context and the adjudication constraints formed by rule priority / coverage relationships jointly determine the final set of compliant rules that take effect.

[0043] The specific process includes, if TransportMode=SEA (sea freight): Activation rule: IF (isInstalledInEquipment == true) AND (hasWh≤100) THENeligibleFor PI967_Section_II [appliesToTransportMode: SEA] Reasoning result: 80Wh≤100Wh, which meets the requirements of PI967 Section II for ocean freight, and can be packaged under these lenient terms.

[0044] If TransportMode = AIR (Air Transport): The aforementioned maritime rules will not be activated because they do not apply to AIR.

[0045] Activation rule: IF (isInstalledInEquipment == true) AND (hasWh>20) THEN NOTeligibleFor PI967_Section_II [appliesToTransportMode: AIR] Reasoning result: 80Wh > 20Wh, which does not meet the requirements of PI967 Section II for air transport, and must be downgraded to the more stringent Section IB or other options.

[0046] It can be seen that the same 80Wh built-in battery is compliant under sea transport (using Section II), but not compliant under air transport (cannot use Section II), which depends entirely on which threshold rule is activated by TransportMode.

[0047] The mandatory requirement for the "Cargo Aircraft Only" (CAO) label applies to the transportation of any lithium batteries (regardless of their Wh capacity) that need to be classified according to UN3481.

[0048] Rule activation and reasoning include: If TransportMode = AIR (Air Transport): The activation rule is: IF (unNumber == “UN3481”) THEN requiresLabel “CargoAircraft Only” [appliesToTransportMode: AIR] The reasoning is that a "Cargo Only" label must be affixed.

[0049] If TransportMode = SEA (sea transport) or ROAD (land transport): The above CAO tag rules will not be activated.

[0050] The activation rule is: IF (unNumber == “UN3481”) THEN requiresLabel “Class 9”, “Lithium Battery Mark” [appliesToTransportMode: SEA, ROAD] The reasoning is that a CAO label is not required, but other general hazardous materials labels are.

[0051] The CAO tag is a mandatory requirement specific to air freight, and it is directly triggered by the condition TransportMode = AIR.

[0052] Regarding the validity requirements of UN38.3 test reports, the scenario is as follows: Although UN38.3 is the basis for all modes of transport, the detailed requirements for its test reports may differ.

[0053] Rule activation and reasoning include: If TransportMode = AIR (Air Transport): The activation rule is: IF (transportMode == AIR) THEN un38.3_testReport mustInclude “T.3 Crush Test” with specificPassCriteria[appliesToTransportMode: AIR] If TransportMode = SEA (Sea Freight): It might activate a more general rule: IF (passed(UN38.3)) THEN eligibleFor UN3481[appliesToTransportMode: SEA], without delving into the details of individual test items.

[0054] Even for the same precondition (UN38.3), the definition of "qualified" may vary slightly under different transport modes, and these differences need to be precisely controlled through context-aware rules.

[0055] In summary, TransportMode, as the meta-condition for rule activation, is key to solving the problem of "one rule, multiple solutions" in UN regulations. It ensures that the output compliance solution is accurately adapted to the current transportation scenario, thereby avoiding compliance risks caused by misuse of rules.

[0056] Taking destination parameter-driven regional constraint rule selection as an example, for air transport scenarios with Germany as the destination, compliance rules with the regional applicability attribute of DE or EU will be added to the candidate compliance rule set; for land transport scenarios with China as the destination, only rules with the regional applicability attribute covering domestic transport scenarios will be activated, and German / EU-specific additional label rules will be filtered out due to mismatch in regional applicability attribute.

[0057] During the rule adjudication phase, rule priority information and coverage relationship information work together on the candidate compliance rule set: rules with higher priority in the same scenario take precedence in case of conflict; coverage relationship information is used to handle situations where special laws take precedence over general laws. For example, if German local regulations have an explicit coverage relationship with some provisions of the IATA general rules, the system will mark the German local regulations as the covering rule during adjudication, so that the inference engine will replace the general IATA requirements with the requirements of the German local regulations in scenarios where the destination is Germany.

[0058] In this implementation, the candidate rule range is formed through the regulatory context, and priority adjudication and coverage resolution are further performed on the candidate rules to keep the rule engine reasoning space consistent with the regulatory retrieval space. This reduces misjudgments caused by the simultaneous triggering of cross-mode, cross-region, and cross-version rules, and reduces the computational overhead caused by irrelevant rules participating in reasoning.

[0059] In the above implementation, preferably, performing forward chain reasoning based on the candidate compliance rule set after adjudication via rule priority information and coverage relationship information includes: When the rule engine performs forward chain reasoning based on the adjudicated candidate compliance rule set, it executes the rules in the order of preset rule groups. The rule groups include at least the eligibility review rule group, the packaging description selection rule group, and the label generation and document requirements rule group.

[0060] The qualification review rule group is used to output transportability conclusions in the compliance judgment results based on the test pass status and prohibition status. Specifically, if UN38.3 is not passed, or the transported object is a damaged battery, defective battery, or recalled battery, the rule engine generates a prohibition on transport and directly outputs a non-transportable conclusion; if the qualification review is passed, the passed fact will continue to be passed to subsequent rule groups.

[0061] The packaging instruction selection rule group is used to output packaging instruction entries based on battery parameters, packaging parameters, and packaging instruction mapping relationships and applicable condition constraints in structured regulatory facts. For example, the rule engine can first determine candidate packaging instruction entries based on parameters such as whether the battery is installed in the device, whether it is in the same box as the device, and its rated watt-hours, combined with the mapping relationship between UN3481 and PI966 and PI967. Then, it can determine the final packaging instruction entry that takes effect in the current scenario by combining the threshold constraint facts, applicable condition constraint facts, and coverage relationship adjudication results corresponding to the current regulatory context.

[0062] The Label Generation and Documentation Requirements Rules group is used to output labeling and documentation requirements based on compliance assessment results, packaging description entries, and clause constraints in structured regulatory facts. For example, when the mode of transport is air freight and the regulatory context meets the relevant clauses, a "Cargo Aircraft Only" label requirement can be output; when there are additional declaration requirements in the destination area or new documentation obligations are added after a specified effective date, additional labels or additional documentation requirements can be added according to the corresponding clause constraints.

[0063] During implementation, a dedicated NLP parser is used to parse normative statements such as "must", "prohibit", and "if..." in the rules and regulations within the UN regulations. Based on the constructed "normative verb-logic operator" mapping table (e.g., "must" → requires; "prohibits If"), thresholds (e.g., "≤100Wh") are automatically extracted and bound to rule variables.

[0064] Among them, UN regulations use a large number of normative verbs with strict logical meanings, which cannot be simply processed by keyword matching, as shown in the table below.

[0065] In this implementation, forward chain reasoning is performed in the order of rule groups: qualification review → packaging instruction selection → label generation and document requirements. This sequentially determines "whether it can be transported," "what packaging to use," and "what labels and documents to attach." This avoids subsequent outputs deviating from the basis of previous judgments, reduces branch drift caused by mutual interference between rules, and improves the stability and regressive verifiability of the output compliance solution. By introducing the mapping relationship and clause constraint information in the structured regulatory facts into the packaging instruction selection rule group and the label generation and document requirements rule group, the rule reasoning results have clause-level traceability support at the factual level.

[0066] In the above implementation, preferably, the compliance scheme includes UN number, packaging instructions, label list and document list, so that the compliance output carries key compliance elements in a structured result, making it easy for the business side to implement according to the list.

[0067] The regulatory basis for tracing the source path includes the triggering rule identifier sequence, the snapshot of the fact object corresponding to the rule identifier sequence, and the regulatory clause number, source document identifier, and effective date information corresponding to the rule identifier sequence. The snapshot of the fact object includes at least the transportation mode parameters, destination parameters, and applicable regulatory time parameters involved in this compliance determination. It may also include parameter values ​​that play a decisive role in this reasoning, such as watt-hours, packaging form, testing status, and prohibition status, to reconstruct the complete input context upon which this compliance determination was based.

[0068] When generating the traceability description text, the system connects the fact object snapshot, regulatory clause number, source document identifier, and regulatory effective date information based on the rule identifier sequence. This organizes the relationships between input facts, triggering rules, corresponding clauses, source documents, effective date, and output requirements into readable text, forming a deliverable traceability description. The traceability description text characterizes the chain of evidence used to obtain a compliance determination result within the current regulatory context (mode of transport, destination, applicable regulatory time).

[0069] During implementation, testing → classification → packaging → labeling → transportation mode form a strictly dependent logical chain. Specifically, it involves UN38.3 (T.1-T.8) → falling under UN3481 → selecting PI966 or PI967 based on "whether it is installed in equipment" → determining the Section level based on watt-hour thresholds and applicable conditions → adjudicating rule coverage based on transportation mode, destination, and applicable regulatory period → outputting final packaging, labeling, and documentation requirements. Failure at any stage of the process breaks the entire logical chain. For example, failure to pass T.5 means UN3481 preconditions cannot be met; or, in a specific destination region and during the effective period of a regulation, if a high-priority rule covers the baseline rule, the system outputs stricter compliance requirements according to the covered rule chain.

[0070] When it is determined that "a certain battery is not transportable", not only is the conclusion output, but also the complete reasoning chain is returned, for example: "Due to failure to pass the T.5 external short circuit test → violation of UN38.3 requirements → failure to meet the premise of UN3481 → transport is prohibited"; when it is determined that "a certain battery can be transported but a higher level of packaging scheme is required", the source explanation can also be output as "Due to the destination being Germany, the mode of transport being air transport, and the applicable time of the regulation being after the supplementary rule came into effect → the regional supplementary rule covers the baseline rule → PI967 Section II is no longer applicable → a higher level of packaging instruction item is used" and the corresponding regulatory clause number, source document identifier and effective time information are attached after each reasoning step.

[0071] In this implementation, by combining rule identifier sequences, fact object snapshots, regulatory clause numbers, source document identifiers, and regulatory effective time information for traceability, compliance conclusions have reproducible basis, reducing the interpretation costs in internal audits, customer reviews, and regulatory verification, and reducing the risk of compliance disputes.

[0072] In the above implementation method, the intelligent compliance method for power battery transportation also includes a feedback closed-loop learning mechanism. When customs rejects a declaration (such as labeling errors), the reason for rejection is automatically captured, the rejection documents are analyzed through NLP, the knowledge graph is located to identify missing or incorrect rules, and repair suggestions are automatically generated (such as adding a new rule: "If the destination is Germany, an additional label X is required").

[0073] During implementation, a multi-agent system (MAS) is used, involving multiple agents, such as: (1) Agent A (Manufacturer): Focuses on the integrity of UN38.3 testing; (2) Agent B (Airline): Focus on CAO labeling and Wh restrictions; (3) Agent C (Customs): Pay attention to the consistency between the declared name and the UN number.

[0074] Each agent operates on the same knowledge graph and its own set of rules, ultimately reaching a consensus. This transforms social compliance conflicts into a multi-agent negotiation mechanism, surpassing traditional single-point decision-making.

[0075] This invention also proposes an intelligent compliance system for power battery transportation, used to implement the intelligent compliance method for power battery transportation disclosed in any of the above embodiments, including: a request information acquisition module, a regulatory context determination module, a regulatory knowledge graph retrieval module, a rule set filtering module, a fact standardization module, a rule engine reasoning module, a compliance output generation module, and a traceability generation module; The request information acquisition module is used to acquire the transportation request information of the power battery to be transported. The transportation request information includes battery parameters, packaging parameters, status parameters, certification document parameters, transportation mode parameters, destination parameters, and regulatory application time parameters used for compliance determination of power battery transportation. The regulatory context determination module is used to determine the regulatory context based on transportation mode parameters, destination parameters, and regulatory application time parameters. The regulatory knowledge graph retrieval module is used to inject the regulatory context as a context-binding variable into the context-aware graph query of the regulatory knowledge graph, and retrieve structured regulatory facts associated with the transport request information from the regulatory knowledge graph under the regulatory context. The structured regulatory facts include clause identifiers, source document identifiers and effective time information corresponding to the relevant regulatory clauses. The rule set filtering module is used to dynamically filter compliance rules based on the regulatory context to obtain a candidate compliance rule set. It then makes a decision on the candidate compliance rule set based on rule priority information and coverage relationship information to obtain a decided candidate compliance rule set, and loads the decided candidate compliance rule set into the rule engine reasoning module. The fact standardization module is used to standardize transport request information and structured regulatory facts into fact objects supported by the rule engine and write them into the rule engine; The rule engine reasoning module is used to perform forward chain reasoning based on the adjudicated candidate compliance rule set, output compliance judgment results and generate a set of compliance actions corresponding to the compliance judgment results; The compliance output generation module is used to generate compliance outputs based on the compliance judgment results and the set of compliance actions. The compliance outputs include the compliance judgment results and the compliance solutions. The compliance solutions include packaging requirements, labeling requirements and documentation requirements. The traceability generation module is used to generate the legal basis traceability path corresponding to the compliance output. The legal basis traceability path includes the triggering rule identifier, the clause identifier associated with the rule identifier, the source document identifier, and the legal effective time information.

[0076] In this implementation, a closed-loop architecture is achieved through a modular approach, encompassing regulatory context determination, context-aware graph retrieval, candidate rule set screening and adjudication, forward chain reasoning, scheme output, and multi-dimensional tracing output, thereby improving system integration controllability and operational consistency.

[0077] In the above embodiments, preferably, the regulatory knowledge graph retrieval module includes a triplet storage submodule and a clause attribute maintenance submodule; The triplet storage submodule is used to store structured regulatory facts in the form of triplets in the regulatory knowledge graph. The clause attribute maintenance submodule is used to configure the source document identifier attribute, clause number attribute, and effective time attribute for each triple, so as to establish a traceable association between the triggered rule identifier and the corresponding legal clause number, source document identifier, and legal effective time information when generating the legal basis tracing path.

[0078] When the source tracing generation module generates the source tracing path based on the regulations, it uses the above attributes to establish the correspondence between the triggered rule identifier and the corresponding regulation clause number, source document identifier, and regulation effective time information, so that the source tracing results have the ability to locate the source document, clause number, and effective period in three dimensions.

[0079] In this implementation, by carrying clause attributes, source attributes, and time attributes at the fact layer, the traceability link is ensured to stably fall at the clause granularity, source granularity, and version granularity, and the efficiency of positioning and comparison in the scenario of regulatory updates is improved.

[0080] In the above embodiments, preferably, the rule set filtering module includes a rule attribute maintenance submodule and a rule filtering submodule; The rule attribute maintenance submodule is used to configure the applicable transportation mode attribute, applicable region attribute, effective time attribute, and rule priority information for each compliance rule, and is also used to maintain the coverage relationship information between compliance rules; The rule filtering submodule is used to dynamically filter compliance rules based on the applicable attributes of the mode of transport, the applicable attributes of the region, and the effective time attributes according to the regulatory context, to obtain a set of candidate compliance rules that match the regulatory context. The submodule then makes a decision on the set of candidate compliance rules based on the rule priority information and the coverage information, to obtain a set of decided candidate compliance rules, and provides the set of decided candidate compliance rules to the rule engine reasoning module.

[0081] In this implementation, the division of labor between rule attribute maintenance and rule filtering enables three-dimensional scenario-based (transportation mode, destination, regulatory version) and on-demand rule loading. Furthermore, the priority and coverage relationship adjudication mechanism resolves rule conflicts during the loading phase, reducing conflicts and performance overhead caused by irrelevant rules participating in reasoning.

[0082] In the above implementation, preferably, the rule engine reasoning module performs forward chain reasoning in the order of preset rule groups, and performs forward chain reasoning based on the candidate compliance rule set after being adjudicated by the candidate rule set screening and adjudication module with rule priority information and coverage relationship information. The rule group includes at least the qualification review rule group, the packaging instruction selection rule group, and the label generation and document requirement rule group. The qualification review rules group is used to output transportability conclusions in compliance judgment results based on test pass and prohibition status; The Packing Description Selection Rule Group is used to output Packing Description entries based on the battery parameters and packaging parameters in the shipping request information, as well as the mapping relationship of Packing Description in the structured regulatory facts and the applicable condition constraints. The Label Generation and Documentation Requirements Rules group is used to output labeling and documentation requirements based on compliance determination results, packaging description entries, and clause constraints in structured regulatory facts.

[0083] In this implementation, by using hierarchical and sequential control of rule groups, the reasoning output is made consistent with the order of business operations, thereby improving the controllability and stability of the reasoning process. By introducing the mapping relationship and clause constraint information in the structured legal facts into the rule groups for selecting packaging instructions and generating labels and document requirements, the legal basis of the reasoning results is supported by the factual layer at the rule group level.

[0084] In the above implementation, preferably, the traceability generation module is used to generate a rule identifier sequence, a snapshot of the fact object corresponding to the rule identifier sequence, and information on the legal clause number, source document identifier, and legal effective time corresponding to the rule identifier sequence. The snapshot of the fact object includes at least the transportation mode parameters, destination parameters, and legal applicable time parameters involved in this compliance determination.

[0085] The source tracing generation module also generates source tracing explanation text based on rule identifier sequences, regulatory clause numbers, source document identifiers, and regulatory effective date information, outputting human-readable source tracing explanations. This source tracing explanation text is used to characterize the chain of evidence for obtaining a compliance determination result within the current regulatory context.

[0086] During implementation, the traceability generation module can organize "Current mode of transport = air, Current destination = DE, Current applicable regulatory time = 2026-03-07, Triggering rule = LBA_2025_DE, Covered rule = IATA_PI967_SecII, Output conclusion = SecIB" into a readable description, so that business personnel, auditors or regulators can directly understand which rule this conclusion comes from, which clause it corresponds to, which regulatory version it is based on, and why other candidate rules are excluded.

[0087] In this implementation, readable traceability instructions are output simultaneously with the delivery of compliance outputs, making the compliance conclusions interpretable, verifiable and auditable, reducing reliance on manual review and improving the transparency of compliance delivery.

[0088] According to the above-described embodiments, the functions to be implemented by each module of the intelligent compliance system for power battery transportation correspond to the steps of the intelligent compliance method for power battery transportation disclosed in the above-described embodiments. During implementation, the above-described embodiments shall be referred to for operation, and will not be repeated here.

[0089] The intelligent compliance method and system for power battery transportation disclosed in the above embodiments are illustrated in the following examples to illustrate the compliance process.

[0090] Example 1: Air transport – Power batteries and equipment are packaged separately (PI967 Section Ⅱ) Scene description: A Chinese new energy vehicle company needs to airlift a batch of lithium iron phosphate (LFP) power battery modules (each with a rated energy of 85Wh) to Germany for after-sales replacement of electric forklifts.

[0091] (1) Each module is not installed in the equipment, but is packed with a forklift controller (i.e., “packed with equipment”). (2) The battery has passed UN38.3 testing and a test summary is attached; (3) The battery state of charge (SoC) is 25%; (4) Each box contains 2 modules, with a total weight of 12kg; (5) The shipper wishes to use IATA PI967 Section II (Exceptions) to be exempt from dangerous goods declaration; (6) Applicable time parameter of regulations: 2026-06-01; Destination parameter: Germany (DE).

[0092] The specific processing flow is as follows: 1. Input parsing User uploads: Battery specifications, UN38.3 report, packaging photos, shipping method (air freight), destination (Germany), applicable date of regulations (2026-06-01).

[0093] The NLP module extracts key parameters: energy = 85Wh, SoC = 25%, and packaging. type = "packed_with_equipment", transport mode = "air", destination = "DE", effective date =“2026-06-01”.

[0094] 2. Regulatory Context Construction and Context-Aware Graph Query The system constructs a three-dimensional regulatory context based on the mode of transport (air), destination (DE), and applicable regulatory date (2026-06-01), and injects it as a context-binding variable into the context-aware graph query, executing the query operation shown in the following figure: (a) Dynamic filtering based on destination scope and effective date: Retrieve regulatory nodes whose scope of application includes DE or EU and whose effective date is earlier than 2026-06-01, while filtering out expired regulatory clauses. (b) Querying attribute value ranges based on battery and packaging parameters: Querying the segment entries in the PI967 node where maxWh≥85, combined with packaging... type =“packed_with_equipment” filters applicable packaging descriptions; (c) Meta-relational reasoning based on rule coverage: detect whether the German local regulatory node has a coverage relationship with the general IATA rules, and if so, attach the coverage relationship information to the candidate compliance rule set.

[0095] In this embodiment, the matching of the PI967 node with the current power battery fact can be calculated using the aforementioned graph node matching degree calculation formula. For the current fact f, where f Wh =85Wh, f Count =2. If the watt-hour interval defined by the regulatory node n corresponding to PI967 Section II is [0,100]Wh and the package quantity interval satisfies that each package does not exceed the specified quantity, then M(n,f)=1, indicating that the regulatory node matches the current power battery facts.

[0096] The system identifies the applicable entry as UN3481, Lithium ion batteries packed with equipment, activates the context as TransportMode=IATA DGR, and loads a subset of PI967 rules.

[0097] 3. Adjudication of the candidate compliance rule set The candidate compliance rule set is adjudicated based on rule priority information and coverage information: When the general IATAPI 967 rule and the German local additional label rule coexist, the German local rule covers the relevant clauses of the general rule in the scenario where the destination is Germany because its geographical applicability is more specific and its rule priority is higher.

[0098] In this embodiment, the candidate rule chain after adjudication can be further sorted using a compliance path potential energy accumulation formula based on graph traversal. Since the region-specific weight of German local additional label rules is higher than that of general IATA rules, and their effective date is closer to the current regulatory application date, compliance paths including German local rules have a higher path potential energy E. path (P), thus being selected by the system as the final rule link to be executed.

[0099] 4. Automated rule-based reasoning (based on Drools rules generated from a knowledge graph) rule “PI967_Section_Ⅱ_Eligibility” when $b: Battery( energy ≤100, soc ≤ 30 ) $p:Packaging(type == “packed_with_equipment”,itemsPerPackage ≤2) $t: Transport( mode == “air” ) then markAsEligibleForSectionⅡ(); requireLabel("Class 9 Lithium Battery Handling Label"); exemptFromShipperDeclaration(); end 5. Compliance Conclusions and Explainable Outputs Conclusion: It meets the conditions of IATA PI967 Section II and can be transported under the exception terms.

[0100] In this embodiment, to quantify the robustness of the conclusion, the system can also calculate the confidence level based on the aforementioned compliance determination confidence formula. Battery Regarding the current facts, if Θ limit =100Wh, f_Wh=85Wh, then the deviation of the physical parameters of the power battery is 0; when the UN38.3 test report is provided, C doc =0, thus a high level of confidence in compliance determination can be obtained to support the output results of transportation that complies with the exception clauses.

[0101] Legal basis and traceability path: Rule identifier sequence: PI967_SectionII_Eligibility → DE_ExtraLabel_Check →SoC_Limit_A154; Fact object snapshot: transport mode =air, destination=DE, effective date =2026-06-01, energy=85Wh, soc=25%, packaging=packed_with_equipment; Source document identification and regulatory effective date: IATA DGR 2025 (effective date: 2025-01-01), PI967, Section Ⅱ, 1.3.2: single battery ≤100Wh; Special Provision A154 (effective date: 2025-01-01): SoC≤30% for air transport; PI967 Section Ⅱ allows a maximum of 2 batteries per pack to be packed with the equipment in the same box.

[0102] Source clarification: Under the regulatory context (air / DE / 2026-06-01), and in accordance with IATADGR 2025 PI967 Section II, Clause 1.3.2 (effective: 2025-01-01) and Special Provision A154, this compliance determination confirms that the battery energy of 85Wh meets the ≤100Wh threshold constraint, the SoC=25% meets the ≤30% limit, and the quantity per box meets the requirement of ≤2 pieces. Therefore, it is determined that transportation is permitted under the exception terms.

[0103] 6. Automatic output (1) Compliance label template (including Class 9 lithium battery symbol + contact number); (2) Packaging Operation Guidelines (PDF); (3) It is not necessary to fill in the "Danger Goods Shipper Declaration".

[0104] Compared to traditional methods, manual review of IATA manuals (Section 967, Clause A154, Section II forms, and German local requirements) can easily lead to overlooking SoCs, quantity limits, or regionally specific requirements. This method can achieve accurate determination within seconds and clearly identify the relevant source documents, clause numbers, and effective dates of regulations through a traceability path based on legal basis. It also supports auditable compliance records and version traceability.

[0105] Example 2: Sea transport – Large-scale power battery integrated into the equipment (PI966 Section I) Scene description: An energy storage system integrator needs to ship a containerized energy storage system (containing a 200 kWh LFP battery cluster) from Shanghai to Australia.

[0106] (1) The battery is permanently installed in the energy storage cabinet (i.e., "contained in equipment"); (2) The battery has passed all UN38.3 tests (T.1-T.8), including vibration, shock, and external short circuit; (3) SoC is 30%; (4) Destination: Australia (AU); Applicable time parameter: 2026-01-01.

[0107] The specific processing flow is as follows: 1. Regulatory Context Construction and Context-Aware Graph Query The system constructs a three-dimensional regulatory context based on the mode of transport (sea), destination (AU), and applicable regulatory date (2026-01-01), and injects context-aware graph queries: dynamically filters and retrieves IMDG regulatory nodes applicable to Australia and effective before 2026-01-01; attribute value range queries confirm that 200kWh exceeds the ≤100Wh threshold limit of PI966 Section II, and routes to Section I; meta-relational reasoning detects whether Australian local maritime rules have any overriding claims against IMDG general rules.

[0108] In this embodiment, the aforementioned graph node matching degree calculation formula can also be used for regulatory node matching in maritime scenarios. Since the current fact corresponds to f Wh The result of matching this node is 0, as it far exceeds the upper limit range corresponding to the node in Section II of PI966. However, for the regulatory node corresponding to Section I of PI966, when its threshold range and applicable conditions are met, the matching function M(n,f)=1, so the system automatically routes the candidate packaging description item to Section I.

[0109] 2. Adjudication of the candidate compliance rule set Following the adjudication of the candidate compliance rule set: PI966 Section I General IMDG rule came into effect; the Australian AMSA additional requirements, due to their territorial applicability attribute (AU) matching the destination and having explicit coverage information, covered some of the General IMDG labeling requirements and were included in the final rule subset.

[0110] In this embodiment, the system calculates the path potential energy E for candidate paths that include the PI966 Section I general IMDG rule and the AMSA additional rule. path (P) Because Australian local supplementary rules are superior in terms of geographical specificity and timeliness, the compliance path that includes AMSA supplementary rules has higher path potential and has been selected as the final compliance path.

[0111] 3. Results of rule-based reasoning Eligibility Review Rules Group: UN38.3 passed in all aspects, no embargo status → can be transported.

[0112] Packaging description selection rule group: contained_in_equipment=true→UN3481, energy=200kWh>100Wh→PI966 Section I (based on packaging description mapping relationships and threshold constraint facts in structured regulatory facts).

[0113] Labeling and Documentation Requirements Rules Group: PI966 Section I + Ocean Freight + Destination AU → Class 9 Labeling + Lithium Battery Marking + IMDG Dangerous Goods Declaration Form + AMSA Additional Documentation Requirements (based on the clause constraints information in the Structured Regulatory Facts).

[0114] 4. Compliance conclusions and legal basis tracing path Conclusion: Transportation is permitted, and shall be operated in accordance with IMDG PI966 Section I.

[0115] In this embodiment, to quantify the robustness of the conclusion allowing transportation, the system can further calculate the confidence level of the compliance determination. Since the key supporting documents are complete in the current scenario, C doc Set to 0; if the watt-hour limit corresponding to the applicable regulatory provisions has not been exceeded, or has been correctly routed to the applicable Section I provisions, then the confidence level of the compliance determination remains at a high level. If a missing UN38.3 report is subsequently detected, then C doc If the maximum value is reached, the confidence level drops rapidly to 0, triggering non-compliant output or manual review.

[0116] Compliance plan: UN number = UN3481, Packaging instructions = PI966 Section I, Label list = [Class 9 label, lithium battery marking], Documentation list = [IMDG Dangerous Goods Shipper's Declaration, UN38.3 Test Report, AMSA Additional Supporting Documents].

[0117] The legal basis for tracing the source includes: rule identifier sequence (UN38.3). Check →UN3481 Classification →PI966_SectionI_Route→AU_AMSA_AdditionalDoc), Fact Object Snapshot (transport) mode = sea, destination = AU, effective date = 2026-01-01, energy = 200kWh, contained_in_equipment = true), regulatory clause number and source document identifier (IMDG Code 2024, PI966, Section I; AMSA Marine Notice 8 / 2025, effective: 2025-07-01) and corresponding regulatory effective date information.

[0118] Source tracing explanation: In the regulatory context (sea / AU / 2026-01-01), this compliance determination was made based on IMDG Code 2024 PI966 Section I (effective: 2024-01-01), confirming that 200kWh exceeds the Section II ≤100Wh threshold, and therefore routed to Section I; in accordance with AMSA Marine Notice 8 / 2025 (effective: 2025-07-01), additional Australian local documentation requirements were imposed; it was determined that the shipment must be declared as a formal dangerous goods shipment and additional AMSA documentation must be submitted.

[0119] 5. Automatic output (1) Dangerous goods transport label (Class 9 + UN3481 + lithium battery mark); (2) Draft IMDG Dangerous Goods Declaration Form (pre-filled IMO Form); (3) MSDS template (including UN38.3 test reference); (4) AMSA Additional Supporting Documents List and Completion Guidelines.

[0120] Compared to traditional solutions, many companies mistakenly believe that if something is installed in equipment, it doesn't need to be declared as dangerous, leading to cargo being detained at ports. This method accurately identifies the additional requirements of the Australian AMSA by identifying the regional constraints in the structured regulatory facts, correctly merges them with the general IMDG rules through the rule coverage relationship mechanism, and clarifies the relevant source document identifiers and the effective date of the regulations through the regulatory basis traceability path, thus avoiding compliance risks caused by the omission of regional special requirements.

[0121] Analysis of Examples 1 and 2 reveals that these two examples fully demonstrate the practical value of the knowledge graph system in cross-border logistics compliance for power batteries: Example 1 accurately activates the PI967 Section II exception clause and identifies regional special requirements through a three-dimensional regulatory context (air freight / Germany / 2026-06-01); Example 2 automatically overlays Australian local supplementary rules through destination parameters (AU) and regional constraint facts. Both resolve rule conflicts through a rule priority and coverage adjudication mechanism, and support the auditability and version traceability of compliance conclusions through a regulatory basis traceability path containing source document identifiers and regulatory effective time information. This prevents over-compliance (increasing costs) and eliminates illegal transportation (causing accidents or fines), truly achieving intelligent, accurate, and reliable regulatory automation.

[0122] Meanwhile, by embedding the graph node matching degree calculation formula, the compliance path potential energy accumulation formula based on graph traversal, and the compliance judgment confidence formula based on power battery attributes into the three key links of regulatory fact retrieval, rule path adjudication, and compliance result output, this invention has clear and calculable technical means in terms of how regulatory facts are matched, how candidate compliance paths are selected, and how compliance results are quantitatively output, thereby further improving the certainty, implementability, and auditability of the technical solution.

[0123] The above are merely preferred embodiments of the present invention and are not intended to limit the present invention. Various modifications and variations can be made to the present invention by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A smart compliance method for transporting power batteries, characterized in that, include: Obtain the transportation request information of the power battery to be transported. The transportation request information includes battery parameters, packaging parameters, status parameters, certification document parameters, transportation mode parameters, destination parameters, and applicable regulatory time parameters for determining the compliance of power battery transportation. A regulatory context is constructed based on the transportation mode parameters, the destination parameters, and the applicable time parameters of the regulations. This regulatory context is then injected as a context-binding variable into a context-aware graph query of the regulatory knowledge graph. Under this regulatory context, structured regulatory facts associated with the transportation request information are retrieved from the regulatory knowledge graph. The context-aware graph query includes at least dynamic filtering based on the destination's applicable scope and effective time, attribute value range queries based on battery parameters and packaging parameters, and meta-relational reasoning based on rule coverage relationships to obtain candidate regulatory facts and determine the coverage relationships in the candidate compliance rule set. Each compliance rule in the candidate compliance rule set carries rule priority information and coverage relationship information. The structured regulatory facts include clause identifiers, source document identifiers, and effective time information corresponding to the corresponding regulatory clauses. The transportation request information and the structured regulatory facts are standardized into fact objects supported by the rule engine. The fact objects are written into the rule engine and forward chain reasoning is performed based on the candidate compliance rule set after adjudication by rule priority information and coverage relationship information. The compliance judgment result is output and a set of compliance actions corresponding to the compliance judgment result is generated. Based on the compliance determination results and the set of compliance actions, compliance outputs are generated. The compliance outputs include compliance determination results and compliance solutions, wherein the compliance solutions include packaging requirements, labeling requirements, and documentation requirements. Output the compliance output and generate a legal basis tracing path corresponding to the compliance output. The legal basis tracing path includes the triggered rule identifier and the clause identifier and source document identifier associated with the rule identifier.

2. The intelligent compliance method for transporting power batteries according to claim 1, characterized in that, The structured regulatory facts are stored in the regulatory knowledge graph in the form of triples, and each triple carries a source document identifier attribute, a clause number attribute, and an effective time attribute. The source document identifier attribute, the clause number attribute, and the effective time attribute are used to establish a traceable association between the triggered rule identifier and the corresponding regulatory clause number, source document identifier, and regulatory effective time information when generating the regulatory basis tracing path. The structured regulatory facts include the correspondence between UN numbers and testing requirements, the mapping between UN numbers and packaging instruction entries, threshold constraint facts and applicable condition constraint facts divided into segmented thresholds and applicable conditions for packaging instruction entries, as well as geographical constraint facts corresponding to the scope of application of the destination and temporal constraint facts corresponding to the period of regulatory effectiveness.

3. The intelligent compliance method for transporting power batteries according to claim 1, characterized in that, The candidate compliance rule set consists of multiple compliance rules. Each compliance rule carries a transportation mode applicability attribute, a geographical applicability attribute, an effective time attribute, rule priority information, and coverage relationship information. The transportation mode applicability attribute is used to indicate the transportation mode to which the compliance rule applies, the geographical applicability attribute is used to indicate the destination area to which the compliance rule applies, and the effective time attribute is used to indicate the applicable time range of the compliance rule. The specific process of constructing a regulatory context based on the transportation mode parameters and obtaining the candidate compliance rule set includes: Based on the transport mode parameters, the regulatory context corresponding to the transport mode is determined, and combined with the destination parameters and the applicable time parameters of the regulations, a regulatory context including the transport mode, destination and applicable time of the regulations is formed; The regulatory context is injected as a context-binding variable into the context-aware graph query of the regulatory knowledge graph to retrieve structured regulatory facts that match the regulatory context. Based on the applicable attributes of the mode of transport, the applicable attributes of the region, and the effective time attributes, the compliance rules are dynamically filtered to obtain a set of candidate compliance rules that match the regulatory context. The candidate compliance rule set is adjudicated based on rule priority information and coverage relationship information to obtain a compliance rule subset, and the compliance rule subset is loaded into the rule engine for the forward chain reasoning.

4. The intelligent compliance method for transporting power batteries according to claim 1, characterized in that, The forward chain reasoning based on the candidate compliance rule set after adjudication via rule priority information and the coverage relationship information includes: performing reasoning in the order of preset rule groups, wherein the rule groups include at least a qualification review rule group, a packaging instruction selection rule group, and a label generation and document requirement rule group; The qualification review rule group is used to output the transportability conclusion in the compliance judgment result based on the test pass status and prohibition status. The packaging description selection rule group is used to output the packaging description item based on the battery parameters and packaging parameters in the transportation request information and the packaging description mapping relationship and applicable condition constraint facts in the structured regulatory facts. The label generation and document requirement rule group is used to output the label requirements and document requirements based on the compliance judgment result, the packaging description item, and the clause constraint information in the structured regulatory facts.

5. The intelligent compliance method for transporting power batteries according to claim 1, characterized in that, The compliance scheme includes UN number, packaging instructions, labeling list, and documentation list; The regulatory tracing path includes a triggered rule identifier sequence, a snapshot of the fact object corresponding to the rule identifier sequence, and the regulatory clause number, source document identifier, and regulatory effective time information corresponding to the rule identifier sequence. The snapshot of the fact object includes at least the transportation mode parameters, destination parameters, and regulatory applicable time parameters involved in this compliance determination. Based on the rule identifier sequence, the regulatory clause number, the source document identifier, and the regulatory effective time information, a tracing explanation text is generated to output a human-readable tracing explanation. The tracing explanation text is used to characterize the chain of evidence for obtaining the compliance determination result in the context of the regulation.

6. A smart compliance system for transporting power batteries, characterized in that, The method for implementing intelligent compliance in the transportation of power batteries as described in any one of claims 1 to 5 includes: a request information acquisition module, a regulatory context determination module, a regulatory knowledge graph retrieval module, a rule set filtering module, a fact standardization module, a rule engine reasoning module, a compliance output generation module, and a traceability generation module; The request information acquisition module is used to acquire the transportation request information of the power battery to be transported. The transportation request information includes battery parameters, packaging parameters, status parameters, certification document parameters, transportation mode parameters, destination parameters, and regulatory application time parameters for determining the compliance of power battery transportation. The regulatory context determination module is used to determine the regulatory context based on the transportation mode parameters, destination parameters, and regulatory applicable time parameters. The regulatory knowledge graph retrieval module is used to inject the regulatory context as a context binding variable into the context-aware graph query of the regulatory knowledge graph, and retrieve structured regulatory facts associated with the transport request information from the regulatory knowledge graph under the regulatory context. The structured regulatory facts include clause identifiers, source document identifiers and effective time information corresponding to the corresponding regulatory clauses. The rule set filtering module is used to dynamically filter compliance rules according to the regulatory context to obtain a candidate compliance rule set, and to adjudicate the candidate compliance rule set according to rule priority information and coverage relationship information to obtain an adjudicated candidate compliance rule set, and to load the adjudicated candidate compliance rule set into the rule engine inference module. The fact standardization module is used to standardize the transportation request information and the structured regulatory facts into fact objects supported by the rule engine and write them into the rule engine. The rule engine reasoning module is used to perform forward chain reasoning based on the adjudicated candidate compliance rule set, output compliance judgment results, and generate a set of compliance actions corresponding to the compliance judgment results; The compliance output generation module is used to generate compliance outputs based on the compliance judgment result and the set of compliance actions. The compliance outputs include the compliance judgment result and the compliance plan. The compliance plan includes packaging requirements, labeling requirements and documentation requirements. The traceability generation module is used to generate a legal basis traceability path corresponding to the compliance output. The legal basis traceability path includes the triggered rule identifier, the clause identifier associated with the rule identifier, the source document identifier, and the legal effective time information.

7. The intelligent compliance system for power battery transportation according to claim 6, characterized in that, The regulatory knowledge graph retrieval module includes a triplet storage submodule and a clause attribute maintenance submodule; The triplet storage submodule is used to store the structured regulatory facts in the form of triples in the regulatory knowledge graph; The clause attribute maintenance submodule is used to configure source document identifier attribute, clause number attribute and effective time attribute for each triple, so as to establish a traceable association between the triggered rule identifier and the corresponding legal clause number, source document identifier and legal effective time information when generating the legal basis tracing path.

8. The intelligent compliance system for power battery transportation according to claim 6, characterized in that, The rule set filtering module includes a rule attribute maintenance submodule and a rule filtering submodule; The rule attribute maintenance submodule is used to configure transportation mode applicable attributes, geographical applicable attributes, effective time attributes, and rule priority information for each compliant rule, and is also used to maintain the coverage relationship information between compliant rules; The rule filtering submodule is used to dynamically filter compliance rules carrying the applicable attributes of the mode of transport, the applicable attributes of the region, and the effective time attribute based on the regulatory context, to obtain a set of candidate compliance rules that match the regulatory context, and to adjudicate the set of candidate compliance rules based on the rule priority information and the coverage relationship information, to obtain a set of adjudicated candidate compliance rules, and to provide the set of adjudicated candidate compliance rules to the rule engine inference module.

9. The intelligent compliance system for power battery transportation according to claim 6, characterized in that, The rule engine reasoning module executes forward chain reasoning in the order of preset rule groups, and executes the forward chain reasoning based on the candidate compliance rule set after being adjudicated by the candidate rule set screening and adjudication module with rule priority information and coverage relationship information. The rule group includes at least the qualification review rule group, the packaging instruction selection rule group, and the label generation and document requirement rule group. The qualification review rule set is used to output transportability conclusions in the compliance judgment results based on the test pass status and prohibition status. The packaging description selection rule group is used to output packaging description entries based on the battery parameters and packaging parameters in the transportation request information, as well as the packaging description mapping relationship and applicable condition constraints in the structured regulatory facts. The label generation and documentation requirements rules group is used to output label and documentation requirements based on the compliance determination results, the packaging instructions, and the clause constraints in the structured regulatory facts.

10. The intelligent compliance system for power battery transportation according to claim 6, characterized in that, The traceability generation module is used to generate a rule identifier sequence, a snapshot of the fact object corresponding to the rule identifier sequence, and information on the regulatory clause number, source document identifier, and regulatory effective time corresponding to the rule identifier sequence. The snapshot of the fact object includes at least the transportation mode parameter, destination parameter, and regulatory applicable time parameter involved in this compliance determination. Based on the rule identifier sequence, the regulatory clause number, the source document identifier, and the regulatory effective time information, the module generates a traceability description text to output a human-readable traceability description. The traceability description text is used to characterize the chain of evidence for obtaining the compliance determination result in the context of the regulation.

Images