A knitting machine triangular configuration intelligent query system and method

Abstract

The application provides a knitting circular machine cam configuration intelligent query system and method, and is applied to the technical field of data processing. Through collecting multi-source data such as cam types and process requirements and parameters such as configuration priority, the application is standardized into a unified query data set through pretreatment; core attributes are extracted and sorted according to parameter importance, and an optimized data set is generated; the search strategy, range and iteration number are determined in combination with the requirement complexity and computing power, a query batch is formed; the efficiency is improved through parallel search, the query algorithm is optimized based on a multidimensional architecture, the strategy is dynamically adjusted through double-index verification; finally, according to the model rule optimization result, redundant schemes are eliminated, and precise, efficient and unique cam configuration intelligent query is realized.

Description

Technical Field

[0001] This invention relates to the field of data processing technology, and in particular to an intelligent query system and method for the triangular configuration of a circular knitting machine. Background Technology

[0002] As a core production equipment in the knitting industry, the cam configuration of the circular knitting machine is a crucial factor determining fabric structure, quality, and production efficiency. Its adaptability directly impacts product qualification rate and production benefits. With the diversification of knitted fabric varieties and the continuous iteration of circular knitting machine models, the parameters of the cam configuration have become increasingly complex. It is necessary to simultaneously match hardware parameters such as cam model, needle track specifications, and needle guide trajectory, as well as process requirements such as knitting density and needle take-off / undocking techniques. Furthermore, it must adapt to characteristic parameters such as fabric raw material composition and elastic modulus, resulting in multi-dimensional and heterogeneous configuration requirements.

[0003] Existing triangular configuration query methods mainly rely on manual experience or simple database retrieval: manual query methods depend on technicians' comprehensive control over equipment parameters, process requirements, and fabric characteristics, resulting in low efficiency, high configuration error rate, reliance on personal experience, and difficulty in large-scale application; simple database retrieval methods can only achieve accurate matching of a single dimension or a small number of parameters, cannot handle the correlation and adaptation of multi-source heterogeneous parameters, and do not consider key factors such as configuration priority and adaptation weight, which easily leads to redundant search results and insufficient accuracy, making it difficult to meet the rapid adaptation needs in complex production scenarios.

[0004] Furthermore, existing query methods lack a standardized processing mechanism for multi-source data, making heterogeneous data integration difficult. This results in a lack of flexibility in retrieval strategies, making it impossible to dynamically adjust based on query complexity and computing resources, further reducing the efficiency and accuracy of configuration queries. Simultaneously, the lack of effective redundancy filtering and accuracy optimization mechanisms in the search results makes it difficult to guarantee the uniqueness and reliability of the output configuration scheme, causing inconvenience in actual production.

[0005] It should be noted that the information disclosed in the background section above is only used to enhance the understanding of the background of this disclosure, and therefore may include information that does not constitute prior art known to those skilled in the art. Summary of the Invention

[0006] Other features and advantages of this application will become apparent from the following detailed description, or may be learned in part by practice of the invention.

[0007] According to one aspect of this application, a method for intelligent querying of cam configurations on a circular knitting machine is provided, comprising: collecting multi-source query data including cam model parameters, knitting process requirements, fabric characteristics, and configuration priority and adaptation weight parameters; performing heterogeneous data standardization transformation through a data preprocessing module to form a query dataset in a unified format; parsing the query dataset, extracting core attributes including parameter dimensions and data matching degree, optimizing data retrieval priority by sorting according to parameter importance, and generating an optimized query dataset that combines parameter completeness and retrieval adaptability; and processing the optimized query dataset in conjunction with query requirement complexity and database computing resources, negotiating and determining a retrieval strategy, and synchronously... The system clearly defines the search scope and matching iteration count, generating query batches containing key parameters and adaptation tags. For data of the same type and with consistent adaptation weights, a parallel retrieval mechanism is employed to improve import efficiency. Based on a multi-dimensional architecture of parameter index engine, process association model, and adaptation score module, parameter mapping is established through association rule learning. Combined with dynamic adaptation weights based on knitting machine type and production scenario, the query algorithm is optimized using weighted matching logic. Simultaneously, adaptation accuracy and response efficiency are verified in real time, and the retrieval strategy is dynamically adjusted in case of anomalies. After query matching is completed, the accuracy of the results is optimized based on machine-specific rules and dynamic adaptation algorithms, and duplicate configuration schemes within the similarity threshold are eliminated through a redundancy filtering mechanism.

[0008] Another aspect of this application is a smart query system for the triangle configuration of a circular knitting machine, comprising:

[0009] The data acquisition module is used to collect multi-source query data, including triangle model parameters, knitting process requirements, and fabric characteristics, and to configure priority and adaptation weight parameters. The data preprocessing module completes the standardization transformation of heterogeneous data to form a query dataset in a unified format.

[0010] The data optimization module is used to parse the query dataset, extract core attributes including parameter dimensions and data matching degree, optimize the data retrieval priority by sorting the parameters by importance, and generate an optimized query dataset that combines parameter completeness and retrieval adaptability.

[0011] The query batch generation module is used to combine the complexity of query requirements with the database computing resources, process the optimized query dataset, negotiate and determine the retrieval strategy, simultaneously clarify the retrieval scope and the number of matching iterations, and generate query batches containing key parameters and adaptation tags.

[0012] The parallel retrieval module is used to employ a parallel retrieval mechanism for data of the same type and with consistent weights, thereby improving data import efficiency.

[0013] The algorithm optimization and retrieval module is used to establish a multi-dimensional architecture based on the parameter index engine, process association model and adaptation scoring module. It establishes parameter mapping through association rule learning, combines knitting machine type and production scenario to dynamically adapt weights, optimizes the query algorithm according to weighted matching logic, and verifies the adaptation accuracy and response efficiency in real time. It also dynamically adjusts the retrieval strategy when anomalies occur.

[0014] The result optimization module is used to optimize the accuracy of the results after the query matching is completed, based on the machine model-specific rules and dynamic adaptation algorithm. It also uses a redundancy filtering mechanism to remove duplicate configuration schemes within the similarity threshold, and outputs triangular configuration query results for large circular knitting machines that are both accurate and unique.

[0015] According to another aspect of this application, an electronic device includes: a first processor; and a memory for storing executable instructions of the first processor; wherein the first processor is configured to execute the above-described intelligent query method for triangular configuration of a circular knitting machine by executing the executable instructions.

[0016] According to another aspect of this application, a computer-readable storage medium is provided having a computer program stored thereon, which, when executed by a second processor, implements the above-described intelligent query method for the triangular configuration of a circular knitting machine.

[0017] This application provides an intelligent query system and method for twill configuration on a circular knitting machine. The aim is to construct a comprehensive solution for multi-source data integration and intelligent retrieval. It collects multi-source data, including twill model parameters, process requirements, fabric characteristics, configuration priority, and adaptation weight parameters. After standardized preprocessing, a unified dataset is formed. Core attributes are extracted and sorted by importance to generate an optimized dataset. The retrieval strategy and iteration count are determined based on requirement complexity and computing power to generate query batches. Parallel retrieval is employed to improve efficiency. Based on a multi-dimensional architecture, parameter mapping is established through association rule learning, dynamically adapting weights and optimizing the query algorithm. Dual-indicator verification is used, and the strategy is adjusted in case of anomalies. Finally, after machine-specific rule verification and redundancy filtering, a precise and unique configuration scheme is output. The entire process integrates multiple algorithms to achieve a closed loop of data processing, retrieval optimization, and result purification, adapting to the needs of complex production scenarios.

[0018] This application addresses the poor adaptability of traditional methods by standardizing multi-source data, quantifying parameter importance, and using a multi-dimensional matching architecture. It also adapts to parallel retrieval mechanisms and dynamic retrieval strategies, reduces interference from redundant data, dynamically adapts weights and adjusts scenario-based strategies, and adapts to the needs of different machine models, processes, and fabrics. Its scope of application covers all types of circular knitting machine production scenarios.

[0019] It should be understood that the above general description and the following detailed description are exemplary and explanatory only, and are not intended to limit this disclosure. Attached Figure Description

[0020] Figure 1 This document illustrates a flowchart of an intelligent query method for the cam configuration of a circular knitting machine according to an embodiment of this application.

[0021] Figure 2 This paper shows a schematic diagram of the structure of an intelligent query system for the triangular configuration of a circular knitting machine according to an embodiment of this application. Detailed Implementation

[0022] The preferred embodiments of the present invention will be described below with reference to the accompanying drawings. It should be understood that the preferred embodiments described herein are for illustration and explanation only and are not intended to limit the present invention.

[0023] The following is combined Figure 1 This application describes an intelligent query method for the cam configuration of a circular knitting machine according to exemplary embodiments thereof. It should be noted that the following application scenarios are shown only to facilitate understanding of the spirit and principles of this application, and the embodiments of this application are not limited in any way. Rather, the embodiments of this application are applicable to any suitable scenario.

[0024] In one implementation, Figure 1 A schematic flowchart illustrating an intelligent query method for the triangular configuration of a circular knitting machine according to an embodiment of this application is shown.

[0025] S101 collects multi-source query data including triangle model parameters, knitting process requirements, and fabric characteristics, and configures priority and adaptation weight parameters. The data preprocessing module completes the standardization transformation of heterogeneous data to form a query dataset in a unified format.

[0026] In one implementation, the collected multi-source query data and parameters need to cover the core business dimensions of cam configuration queries for circular knitting machines, specifically as follows: Cam model parameters include cam model codes, guide needle trajectory curve parameters, number and spacing of needle tracks, cam material hardness parameters, and mounting hole dimensions, comprehensively reflecting the structural and functional characteristics of the cam. Knitting process requirements include key process execution indicators such as fabric weave type (plain weave, rib weave, jacquard, etc.), knitting density parameters, loop size specifications, needle take-up and take-down process requirements, and knitting speed adaptation range. Fabric characteristic data involves core fabric attributes affecting cam adaptation, such as fabric raw material composition (cotton, polyester, spandex, etc. and their proportions), fabric weight, yarn fineness, fabric elastic modulus, and tensile strength.

[0027] Based on production needs, clear prioritization criteria are established, such as prioritizing process precision, compatibility with existing equipment, and cost control. These criteria are set based on industry practices and production data, such as a weighting of 0.4 for triangular model parameter adaptation, 0.3 for process requirement adaptation, and 0.3 for fabric characteristic adaptation. These can be dynamically adjusted according to actual application scenarios.

[0028] A dedicated data preprocessing module is used to standardize the conversion of heterogeneous data, ensuring that the data format is uniform and the dimensions are consistent. The specific processing flow is as follows: unstructured data (such as process document text), semi-structured data (such as Excel spreadsheet records), and structured data (such as database storage parameters) from different sources are uniformly converted into JSON format, and the field naming conventions and data types (numeric, character, enumeration, etc.) are clearly defined. For example, the triangular model code is uniformly unified into a fixed-length string of "letters + numbers", and numeric parameters are uniformly retained to two decimal places.

[0029] Missing data is handled using a missing value imputation algorithm (numerical parameters are imputed with the mean, and enumerated parameters with the mode). An outlier detection algorithm (based on the 3σ principle) identifies and removes outliers exceeding reasonable ranges, such as outliers in the number of triangular stitches or unusual fabric weights. Dimensions from different data sources are standardized, missing dimension information is added, and redundant dimensions are removed to ensure all data contains the core dimensions required for triangular configuration queries, forming a standard set of data fields.

[0030] The Min-Max standardization method is used to map numerical parameters to the [0,1] interval, eliminating the impact of differences in units. For example, parameters of different magnitudes, such as weaving speed and fabric weight, are standardized in a unified manner. The final output is a query dataset with a unified format, consistent dimensions, and clean data, providing high-quality data support for the subsequent retrieval process.

[0031] S102, parse the query dataset, extract core attributes including parameter dimensions and data matching degree, optimize data retrieval priority by sorting by parameter importance, and generate an optimized query dataset that combines parameter completeness and retrieval adaptability.

[0032] In one implementation, to address the characteristics of the query dataset containing multi-source heterogeneous parameters and differences in core attribute dimensions, a full-process processing mechanism is established, encompassing data parsing, core attribute extraction, parameter importance quantification, and retrieval priority ranking, to achieve data normalization. Then, through a hierarchical optimization strategy of assigning weights to key parameters and filtering secondary parameters, and step-by-step execution of importance scoring and priority ranking, a basic set of query data for retrieval covering all elements from attributes to weights to ranking and retrieval adaptation is formed. To address the multi-source heterogeneity and differences in core attribute dimensions of the query dataset, a four-step full-process processing mechanism is established, simultaneously employing hierarchical optimization and step-by-step execution strategies to form the basic set of query data for retrieval: A JSONSchema parsing algorithm and an XML / CSV structured conversion algorithm are used to parse the standardized query dataset. The JSONSchema algorithm defines the type constraints, format specifications, and relationships of data fields, deconstructing nested JSON structures; the XML / CSV conversion algorithm flattens non-JSON format data into a one-dimensional key-value pair structure, ensuring hierarchical consistency across different data sources, allowing direct calls from subsequent modules.

[0033] Based on the triangular configuration query business logic of the circular knitting machine, two core attributes are extracted: parameter dimension and data matching degree. The parameter dimension is determined by the field enumeration algorithm to clarify the basic characteristics of each data, such as the number of fields, the range of values, and the data type. The data matching degree is calculated by the cosine similarity algorithm. By constructing the feature vectors of the query data and the historical valid configuration data, the cosine value of the angle between the vectors is solved to quantify the data adaptation potential. The value range is [0,1].

[0034] A combined weighting algorithm using the Analytic Hierarchy Process (AHP) and entropy weighting is employed. First, an AHP algorithm is used to construct a three-tiered hierarchical structure: target layer (query result accuracy), criteria layer (triangle model, process requirements, fabric characteristics), and solution layer (specific parameters). Five to eight industry technical experts are invited to conduct pairwise comparisons of elements at each level, constructing a judgment matrix and calculating subjective weights. Then, the entropy weighting method is used to calculate the information entropy and difference coefficient of each parameter, obtaining objective weights. Finally, the parameters are weighted and merged at a ratio of 0.6 for subjective weights and 0.4 for objective weights to obtain a comprehensive importance score for each parameter.

[0035] The quicksort algorithm is used to sort the query data in descending order, with the overall importance score of the parameters as the sorting criterion. Parameters with higher scores have higher retrieval priority, and the sorting time complexity is controlled to O(nlogn), ensuring efficient processing of large-scale datasets. For the top 30% of key parameters in terms of importance score, a fixed weighting method is used to assign weight values ​​of 0.7-1.0 to enhance their retrieval influence. For the bottom 20% of secondary parameters, a variance filtering algorithm is used to calculate the parameter variance, eliminating low-discrimination parameters with variance less than 0.01 to reduce the retrieval load.

[0036] First, the importance score of all data is completed by combining weighting algorithms. Then, the quicksort algorithm is called to perform priority sorting to ensure that the sorting logic is consistent with the scoring criteria. Finally, a basic set of full-element query data retrieval covering attribute definition, weight allocation, priority sorting, and retrieval adaptation rules is formed.

[0037] To address the core requirements of accurate parameters and efficient retrieval for triangular configuration queries on circular knitting machines, a query data priority optimization mechanism is constructed. This mechanism includes a parameter importance assessment module to precisely analyze and refine the retrieval needs and priority ranking criteria for different types of query data. The complete priority optimization mechanism, centered on parameter importance assessment, links the data parsing module, sorting execution module, and verification and adjustment module, forming a closed-loop logic of "assessment-sorting-verification." The data parsing module outputs structured data, the sorting execution module completes the sorting based on the assessment results, and the verification and adjustment module reports sorting anomalies, ensuring a high degree of alignment between priority settings and query requirements.

[0038] The module incorporates a business rule base and a combined weighting algorithm model. The business rule base covers key influencing factors of triangle configuration in knitting production, such as "prioritizing triangle guide needle trajectory parameters when querying jacquard processes" and "strengthening the weight of triangle material hardness parameters when querying elastic fabrics." The algorithm model adopts the aforementioned AHP-entropy weighting algorithm, which clarifies the priority ranking criteria by sorting out the retrieval needs of different types of query data, and achieves a combination of objective and subjective importance scoring.

[0039] A parameter attribute-retrieval priority mapping rule is constructed to transform the core feature requirements of the query data into executable sorting logic. A parameter matching degree threshold is introduced to standardize and validate the sorting results. An anomaly data secondary screening mechanism adjusts substandard sorting, generating an optimized query dataset that balances parameter completeness and retrieval adaptability. Based on the parameter importance assessment results, a decision tree algorithm is used to construct the parameter attribute-retrieval priority mapping rule. Using parameter type (critical / minor), importance score range, and data matching degree as decision nodes, a classification rule tree is generated using the C4.5 algorithm, transforming the core feature requirements of the query data into executable sorting logic. For example, query data with critical parameters and scores ≥0.8 and matching degrees ≥0.7 are directly assigned the highest retrieval priority; query data with minor parameters and scores ≤0.3 and matching degrees <0.5 are assigned a basic retrieval priority.

[0040] A parameter matching threshold (default 85%) is introduced, and a threshold comparison algorithm is used to validate the sorted query data. Sorting results with a matching degree below the threshold are considered substandard and trigger a secondary processing flow. An isolated forest algorithm is used to identify outlier data. Multiple isolated trees are constructed to cluster the substandard sorted data, and samples deviating from normal clusters are identified as anomalous. For anomalous data, a parameter completion algorithm is used to supplement missing key parameters, a linear interpolation algorithm is used to correct data with inconsistent dimensions, a cosine similarity algorithm is recalculated to determine the matching degree, and finally, invalid data that cannot be corrected is removed.

[0041] After executing decision tree mapping rules, threshold comparison and verification, and secondary screening by isolated forest, an optimized query dataset with both parameter completeness and retrieval adaptability is generated, ensuring that the data can be directly used for subsequent retrieval strategy formulation and query batch generation.

[0042] S103, combining the complexity of query requirements with the computing power resources of the database, processes the optimized query dataset, negotiates and determines the retrieval strategy, simultaneously clarifies the retrieval scope and the number of matching iterations, and generates query batches containing key parameters and adaptation tags.

[0043] In one implementation, a query requirement complexity analysis is conducted on the optimized query dataset to identify core indicators such as the process adaptation accuracy threshold, parameter matching constraints, and fabric characteristic adaptation difficulty. These indicators are then categorized and labeled according to their complexity level. A multi-dimensional index quantification method is employed to define and quantify the three core indicators. The process adaptation accuracy threshold is defined by the allowable deviation range, meaning the parameter deviation between the actual configuration and the required process does not exceed ±2%. Parameter matching constraints are determined using a constraint item counting method, counting the number of parameters that must be strictly matched (such as triangle patterns, number of stitches, etc.). Fabric characteristic adaptation difficulty is assessed using a hierarchical evaluation method, classifying the difficulty coefficient according to fabric composition complexity, elastic modulus range, and tensile strength level, with values ​​ranging from 0 to 1.

[0044] K-means clustering algorithm was used to perform cluster analysis on three types of core indicator data, classifying them into low, medium, and high complexity levels. Low complexity: allowable deviation of process adaptation accuracy ≥ ±2%, number of constraint parameters ≤ 3, fabric adaptation difficulty coefficient ≤ 0.3; Medium complexity: allowable deviation of process adaptation accuracy ±1%-±2%, number of constraint parameters 4-6, fabric adaptation difficulty coefficient 0.3-0.7; High complexity: allowable deviation of process adaptation accuracy ≤ ±1%, number of constraint parameters ≥ 7, fabric adaptation difficulty coefficient ≥ 0.7. This algorithm was used to label the complexity level of all query requirements, ensuring objective classification.

[0045] A quantitative assessment of database computing resources was conducted to obtain key performance parameters such as data processing throughput, retrieval response latency, and concurrent processing capability, and a computing resource supply and demand matching evaluation table was established. A performance benchmarking algorithm was employed to obtain the three key performance parameters by simulating concurrent execution of query tasks of different scales. Data processing throughput was calculated by the number of data matches completed per unit time, in rows per second; retrieval response latency was calculated by recording the total time taken from query initiation to result return, and taking the average of multiple tests; concurrent processing capability was measured by gradually increasing the number of concurrent query threads, recording the maximum number of threads during stable system operation.

[0046] Based on the above quantitative parameters, a computing power resource supply and demand matching evaluation table is constructed, clarifying the numerical range, level classification, and corresponding query complexity of each parameter. For example, data processing throughput ≥ 1000 records / second, response latency ≤ 500 milliseconds, and concurrent threads ≥ 50 are classified as high-level computing power, suitable for high-complexity query requirements; throughput 500-1000 records / second, latency 500-1000 milliseconds, and concurrent threads 30-50 are classified as medium-level computing power, suitable for medium-complexity requirements; throughput < 500 records / second, latency > 1000 milliseconds, and concurrent threads < 30 are classified as low-level computing power, suitable for low-complexity requirements.

[0047] Based on the classification results of demand complexity and the assessment data of computing resources, a suitable retrieval strategy is determined through negotiation using a supply-demand balancing algorithm. High-complexity demands are matched with a precise retrieval strategy, while medium-to-low complexity demands are matched with a fast retrieval strategy. A supply-demand balancing model is constructed using a linear weighted summation method. The three core indicators of demand complexity and the three key parameters of computing resources are assigned weights (complexity indicators have a total weight of 0.5, and computing resource parameters have a total weight of 0.5), and a supply-demand matching score is calculated. A score ≥ 0.8 indicates a high degree of supply-demand matching, and a precise retrieval strategy is adopted; a score between 0.5 and 0.8 indicates a moderate match, and a fast retrieval strategy is adopted; a score < 0.5 triggers computing resource scheduling optimization, prioritizing core query demands.

[0048] The precise retrieval strategy employs a full-parameter traversal matching logic, comparing all parameters such as triangle model, process requirements, and fabric characteristics one by one to ensure matching accuracy. The fast retrieval strategy employs a core parameter priority matching logic, comparing only the top 50% of core parameters by importance score, while secondary parameters use fuzzy matching rules to improve retrieval efficiency.

[0049] By combining the parameter size, parameter correlation, and matching difficulty of the optimized query dataset, and adhering to the principle of full coverage of core parameters and selective filtering of secondary parameters as needed, the scope and boundaries of the retrieved data are clearly defined. The Apriori algorithm (an association rule mining algorithm) is used to analyze the parameter size, parameter correlation, and matching difficulty of the optimized query dataset. Parameter size is measured by the total number of fields, parameter correlation is determined by calculating the support and confidence between each parameter, and matching difficulty is quantified by combining historical matching success rates.

[0050] Following the principle of full coverage of core parameters and selective filtering of secondary parameters as needed, the scope of data retrieval is clearly defined based on association rules mined using the Apriori algorithm. Core parameters (such as triangular model codes, fabric weave types, and fabric raw material composition) are all included in the retrieval scope. Among secondary parameters, those with a correlation support of ≥0.6 with the core parameters are included in the retrieval scope; those with a correlation support below 0.6 are dynamically filtered based on matching difficulty, with those having a matching difficulty coefficient ≥0.5 included, otherwise removed. This defines the retrieval boundaries and reduces the participation of invalid data in the matching process.

[0051] Based on the selected retrieval strategy and data range, and considering the upper limit of computing power, the number of iterations for target matching is determined through iterative efficiency simulation. A Monte Carlo simulation algorithm is used to simulate the matching process under different retrieval strategies and data ranges. Using retrieval accuracy and execution efficiency as objective functions, parameters such as retrieval strategy type, data range size, and upper limit of computing power are input, and multiple rounds of simulation iterations are performed, recording the objective function values ​​at different iteration counts. Based on the simulation results, the number of iterations when the objective function is optimal is selected as the target value. For high-complexity requirements using a precise retrieval strategy, the number of iterations is set to 3-5 to ensure comprehensive matching; for medium-complexity requirements using a fast retrieval strategy, the number of iterations is set to 2-3; and for low-complexity requirements using a fast retrieval strategy, the number of iterations is set to 1-2 to achieve a balance between matching effectiveness and efficiency.

[0052] Key parameters and adaptation tags are extracted from the optimized query dataset, and retrieval execution rules are integrated to generate structured query batches that can be directly used for retrieval. A field filtering algorithm is used to extract key parameters and adaptation tags from the optimized query dataset. Key parameters include triangle model, process adaptation accuracy threshold, and core fabric characteristics; adaptation tags include complexity level, computing power adaptation level, and retrieval strategy type.

[0053] The execution logic, parameter comparison order, and matching threshold standards corresponding to the retrieval strategy are analyzed and integrated with the extracted key parameters and adaptation tags to generate structured query batches in JSON format. Each query batch contains fields such as batch identifier, key parameter set, adaptation tag set, retrieval execution rules, and iteration count, and can be directly called and executed by the retrieval module.

[0054] S104. A parallel retrieval mechanism is used to improve import efficiency for data of the same type and with consistent adaptation weights.

[0055] In one implementation, a K-nearest neighbor classification algorithm is used. Based on core features such as triangular model parameter category, knitting process requirement type, and fabric characteristic attribution, the optimized query dataset is clustered and grouped. Data with feature similarity ≥ 0.85 are classified into the same type, ensuring that data in the same group have consistent retrieval adaptation scenarios. The Euclidean distance algorithm is used to calculate the adaptation weight difference between data of the same type. A weight difference threshold of 0.05 is set. When the Euclidean distance between all pairs of data in the same group is ≤ 0.05, it is determined that the adaptation weight is consistent, forming a data group that can be used for parallel retrieval. If there is data with a weight difference exceeding the threshold, it is split into groups with corresponding weights, ensuring that the data type and adaptation weight of each group are consistent.

[0056] A three-tiered architecture is constructed, consisting of a data grouping and scheduling module, a multi-threaded retrieval and execution module, and a result aggregation and integration module. The data grouping and scheduling module receives qualified data groups and assigns retrieval tasks to each group. The multi-threaded retrieval and execution module has a built-in thread pool that dynamically allocates thread resources based on the size of the data groups. The result aggregation and integration module collects the retrieval results from each thread and outputs them in a standardized format. The core thread count of the thread pool is set to 1.5 times the number of CPU cores in the database, and the maximum number of threads does not exceed twice the number of CPU cores to avoid resource contention. The thread queue capacity is configured to twice the total number of data groups retrieved in a single operation, and the idle thread lifespan is set to 60 seconds to ensure efficient reuse of thread resources.

[0057] The data grouping and scheduling module employs a round-robin scheduling algorithm to evenly distribute each data group among idle threads in the thread pool, ensuring balanced load on each thread. During allocation, the mapping relationship between data group identifiers and thread IDs is recorded for easy result traceability. Each thread synchronously starts its retrieval task. Based on the distributed indexing mechanism of the parameter indexing engine, data groups of the same type and weight quickly locate data resources through shared index partitions, avoiding duplicate index queries. A non-blocking I / O model is used during retrieval, allowing threads to start processing the next data without waiting for the previous data retrieval to complete, improving the parallelism of import and retrieval. After each thread completes its retrieval, it temporarily stores the results in a local cache and synchronously sends a completion signal to the result aggregation and integration module. Upon receiving completion signals from all threads, the module uses a merge sort algorithm to integrate the temporarily stored results from each thread in descending order of their suitability scores, generating a unified retrieval result set and ensuring consistent result order.

[0058] An optimistic locking mechanism is employed to avoid conflicts caused by multiple threads accessing the same data resource simultaneously. A temporary marker is added when a thread accesses data, and the marker is released after access is complete. If the marker is occupied, the thread waits 10 milliseconds before retrying, with a maximum of 3 retries to ensure stable retrieval. A load monitoring algorithm is used to monitor the CPU utilization, memory usage, and retrieval response time of each thread in real time. When a thread's load is ≥80%, unfinished tasks are split and moved to idle threads. When the overall thread pool load is ≤30%, the number of core threads is appropriately reduced to achieve dynamic load balancing and ensure a stable improvement in import efficiency.

[0059] S105 is based on a multi-dimensional architecture of parameter indexing engine + process association model + adaptability scoring module. It establishes parameter mapping through association rule learning, combines knitting machine type and production scenario to dynamically adapt weights, optimizes query algorithm according to weighted matching logic, and verifies adaptation accuracy and response efficiency in real time. It also dynamically adjusts search strategy when anomalies occur.

[0060] In one implementation, a multi-dimensional architecture based on a parameter indexing engine, a process association model, and a fit scoring module is used. This architecture, combined with knitting machine models and production scenario requirements, establishes a multi-dimensional parameter mapping between triangular parameters, processes, and fabrics through association rule learning, determining the dynamic fit weight allocation logic. The multi-dimensional architecture consists of a parameter indexing engine, a process association model, and a fit scoring module connected in series. The parameter indexing engine, as the data entry point, is responsible for quickly locating the associated data of triangular parameters, process requirements, and fabric characteristics. The process association model receives the indexed data and establishes the fit logic among the three. The fit scoring module quantifies the matching results and outputs a fit score, forming a closed-loop chain of "data indexing - logical association - score output".

[0061] The Apriori association rule algorithm is used, with historical successful triangulation cases as training data, to mine frequent itemsets and strong association rules between triangulation parameters (such as the number of stitches), process requirements (such as weaving density), and fabric characteristics (such as yarn fineness). The support threshold is set to 0.6 and the confidence threshold is set to 0.75 to generate a multi-dimensional parameter mapping table and clarify the adaptation relationship of different parameter combinations.

[0062] Based on the requirements of knitting machine type (such as single-sided / double-sided circular knitting machine) and production scenario (such as mass production / sample trial production), a fuzzy comprehensive evaluation method is used to determine the weight allocation logic. The weights are set as follows: machine type adaptation weight 0.4, scenario adaptation weight 0.3, and historical adaptation success rate weight 0.3. The dynamic adaptation weights of each parameter are calculated using a fuzzy matrix, with weight values ​​ranging from 0 to 1, and are adjusted in real time according to changes in the type and scenario.

[0063] Based on the parameter importance ranking results and retrieval strategy requirements, the query algorithm is optimized using a weighted matching logic, clarifying the execution logic of prioritizing core parameter comparison and gradient matching of secondary parameters. Based on the parameter importance ranking results and retrieval strategy (accuracy / speed), a weighted cosine similarity algorithm is used to optimize the query algorithm. The weight coefficients for core parameters (top 50% of importance scores) are set to 1.0-1.5, and the weight coefficients for secondary parameters (bottom 50% of importance scores) are set to 0.5-1.0. The similarity between the query data and candidate configuration data is calculated using weighted averages; the higher the similarity value, the higher the matching priority.

[0064] Define the execution rules for prioritizing the comparison of core parameters and then performing gradient matching of secondary parameters. First, perform a full and accurate comparison of core parameters such as triangular model codes and fabric structure types to filter out candidate data with a core parameter matching degree of ≥90%. Then, perform fuzzy comparison of secondary parameters according to their importance gradient, successively verifying fabric weight, weaving speed compatibility range, etc., to gradually narrow down the range of candidate data and improve matching efficiency.

[0065] A real-time verification mechanism based on two indicators—adaptation accuracy and response efficiency—is employed. An accuracy threshold and a response efficiency upper limit are set, and changes in these indicators are continuously monitored throughout the query process. Adaptation accuracy is the percentage of configuration schemes in the matching results that meet the process and model requirements, with a pass threshold set at 90%. Response efficiency is the total time taken for a single query from initiation to return a result, with an upper limit set at 3 seconds. Both indicators are hard constraints and must be met simultaneously. A sliding window monitoring algorithm is used, with 10 queries per monitoring window. Adaptation accuracy and response efficiency data for each query are collected in real time, and the average and standard deviation within the window are calculated to dynamically track indicator trends and prevent single instances of abnormal data from influencing the judgment.

[0066] If the adaptation accuracy fails to meet the target, a parameter weight recalibration process is triggered. When the adaptation accuracy within a window falls below 90%, the parameter weight recalibration process is triggered. A particle swarm optimization algorithm is used, aiming to improve the accuracy above a threshold. The weight coefficients of core and secondary parameters are iteratively optimized, with 50 iterations, a learning factor of 1.5, and linearly decreasing inertia weights, until the weight configuration meets the accuracy requirements.

[0067] In response to timeout information, a retrieval strategy downgrade is initiated. When a query takes more than 3 seconds, the retrieval strategy downgrade is activated. The precise retrieval strategy is downgraded to a fast retrieval strategy, the gradient verification step of fuzzy comparison of secondary parameters is disabled, and only the full comparison of core parameters and the verification of highly correlated secondary parameters are retained; at the same time, the retrieval depth of the parameter index engine is adjusted from full index query to core field index query, shortening the data location time.

[0068] The adjusted query algorithm and retrieval strategy undergo secondary validation, with parameter adjustment logs and strategy optimization trajectories recorded synchronously to form a stable query execution plan that balances matching accuracy and execution efficiency. Cross-validation is employed, using 20% ​​of the query data as a test set to perform secondary validation on the adjusted query algorithm and retrieval strategy. This also verifies whether the adaptation accuracy and response efficiency meet the threshold requirements. If they still do not meet the standards, the exception handling process is repeated until both metrics stably meet the requirements. Information such as the values ​​of parameter weights before and after adjustment, the content of retrieval strategy adjustments, and validation results are recorded synchronously to form a strategy optimization trajectory archive. Once the metrics stably meet the requirements for three consecutive monitoring windows, the current algorithm parameter configuration and retrieval strategy rules are solidified as a standard execution plan for use in subsequent similar query tasks.

[0069] S106 After completing the query matching, the accuracy of the results is optimized based on the model-specific rules and dynamic adaptation algorithm, and duplicate configuration schemes within the similarity threshold are eliminated through a redundancy filtering mechanism.

[0070] In one implementation, the model compatibility requirements, parameter matching completeness, and process compatibility consistency in the query matching results are parsed and standardized to generate model-specific verification conditions, core parameter matching thresholds, and process compatibility qualification standards, forming the basic information for optimizing result accuracy. A semantic rule parsing algorithm (based on Prolog logic programming) is used to perform structured parsing of the model compatibility requirements, parameter matching completeness, and process compatibility consistency. By defining a fact base (e.g., "Model M1 is compatible with 3-6 needle channels") and a rule base (e.g., "Missing core parameters result in incomplete matching"), natural language rules are transformed into logical expressions, achieving automated parsing and standardized transformation.

[0071] Model-specific verification conditions are clearly constrained by an interval judgment algorithm, such as limiting the distance between the triangular mounting holes of a single-sided large circular knitting machine to 15-20mm; the core parameter matching threshold adopts a binary matching algorithm, requiring the binary bit matching degree of core parameters such as triangular model code and guide needle trajectory parameters to be ≥95%; the process adaptability qualification standard is set by a deviation rate calculation algorithm, with weaving density deviation ≤±3% and coil size error ≤±0.2mm, forming basic information.

[0072] The parameter adjustment logic, result correction range, and accuracy improvement target of the dynamic adaptation algorithm are constrained, defined, and quantified to generate adaptation parameter optimization coefficients, result correction boundaries, and accuracy achievement thresholds, forming result optimization constraint information. A linear constraint programming algorithm is used to define the parameter adjustment logic (e.g., the adjustment magnitude of core parameter weights is positively correlated with accuracy improvement), the result correction range (e.g., trigonometric parameter correction does not exceed ±5% of the design value), and the accuracy improvement target (e.g., improving adaptation accuracy from 88% to 92%), transforming qualitative requirements into quantitative constraint equations.

[0073] The optimization coefficients of the adaptation parameters are solved iteratively using the gradient descent algorithm. The optimization coefficients of the core parameters are set to 1.0-1.2, and the coefficients of the secondary parameters are set to 0.8-1.0. The result correction boundary is limited by a threshold truncation algorithm. Correction values ​​exceeding ±5% are directly truncated as boundary values. The accuracy threshold is set by a statistical sampling algorithm. Based on 1000 sets of historical valid data, the adaptation accuracy rate is determined to be ≥92% as the standard, forming constraint information.

[0074] The criteria for judging redundant configuration schemes, the dimensions of similarity calculation, and the execution logic of filtering are refined and parameterized to generate duplicate scheme judgment conditions, multi-dimensional similarity calculation benchmarks, and redundancy filtering execution thresholds, forming redundancy removal execution information. The FP-Growth frequent pattern mining algorithm is used to mine frequent itemsets of redundant schemes (such as duplicate configurations of "triangle model A + plain weave process + cotton fabric") from 100,000 historical query data sets, refining redundancy judgment rules and transforming them into parameterized logic. Duplicate scheme judgment conditions are set using a multi-dimensional weighted matching algorithm; a core parameter matching degree ≥90% and a secondary parameter matching degree ≥85% are considered suspected redundancy. Multi-dimensional similarity calculation uses an improved cosine similarity algorithm, assigning weights of 0.4 to the triangular parameters, 0.3 to the process requirements, and 0.3 to the fabric characteristics to calculate the comprehensive similarity. The redundancy filtering execution threshold is determined using a ROC curve analysis algorithm, taking 92% of the similarity value corresponding to a false positive rate ≤5% as the threshold, forming removal execution information.

[0075] This system integrates basic information for result accuracy optimization, result optimization constraints, and redundancy removal execution information. It performs end-to-end collaborative processing of query result accuracy verification, dynamic adaptation optimization, and redundant scheme removal to generate a final query result for the triangular configuration of a circular knitting machine that combines accuracy and uniqueness. Employing a data fusion algorithm (DS evidence theory), it integrates credibility factors from the three types of information to construct a collaborative decision-making model of "accuracy verification - dynamic adaptation optimization - redundancy removal." The output of each stage serves as input evidence for the next, achieving a closed-loop process.

[0076] First, a rule-matching algorithm is used to perform accuracy verification, eliminating schemes that do not meet the model requirements, fail to match core parameters, or are not compatible with the process. Then, a particle swarm optimization algorithm is used for dynamic adaptation optimization, adjusting parameter weights and correcting data deviations to ensure that the accuracy threshold is met. Finally, a similarity clustering algorithm (DBSCAN) is used to cluster schemes with a comprehensive similarity of ≥92% into redundant clusters, retaining the scheme with the highest adaptation within the cluster and eliminating the rest of the redundancy.

[0077] After collaborative processing, the final query results are generated, including triangular model number, process parameters, fabric matching suggestions, and matching score. The data is output with a unified structure through a JSON formatting algorithm and can be directly used for production configuration.

[0078] In one implementation, such as Figure 2 As shown, this application also provides an intelligent query system for the triangle configuration of a circular knitting machine, including:

[0079] The data acquisition module 201 is used to collect multi-source query data including triangle model parameters, knitting process requirements, and fabric characteristics, as well as configure priority and adaptation weight parameters. The data preprocessing module completes the standardization transformation of heterogeneous data to form a query dataset in a unified format.

[0080] Data optimization module 202 is used to parse the query dataset, extract core attributes including parameter dimensions and data matching degree, optimize data retrieval priority by sorting by parameter importance, and generate an optimized query dataset that combines parameter completeness and retrieval adaptability.

[0081] The query batch generation module 203 is used to combine the complexity of query requirements with the computing power resources of the database, process the optimized query dataset, negotiate and determine the retrieval strategy, simultaneously clarify the retrieval scope and the number of matching iterations, and generate query batches containing key parameters and adaptation tags.

[0082] Parallel retrieval module 204 is used to employ a parallel retrieval mechanism for data of the same type and with consistent adaptation weights, thereby improving data import efficiency.

[0083] The algorithm optimization and retrieval module 205 is used to establish a multi-dimensional architecture based on the parameter index engine, process association model and adaptation scoring module. It establishes parameter mapping through association rule learning, combines knitting machine type and production scenario to dynamically adapt weights, optimizes the query algorithm according to weighted matching logic, and verifies the adaptation accuracy and response efficiency in real time. It also dynamically adjusts the retrieval strategy when anomalies occur.

[0084] The result optimization module 206 is used to optimize the accuracy of the results after the query matching is completed, based on the machine type specific rules and dynamic adaptation algorithm. It also uses a redundancy filtering mechanism to remove duplicate configuration schemes within the similarity threshold, and outputs a triangular configuration query result for the large circular knitting machine that is both accurate and unique.

[0085] The computer-readable storage medium provided in the above embodiments of this application and the intelligent query method for the triangular configuration of the circular knitting machine provided in the embodiments of this application are based on the same inventive concept and have the same beneficial effects as the methods adopted, run or implemented by the application programs stored therein.

[0086] The various embodiments in this application are described in a related manner. Similar or identical parts between embodiments can be referred to mutually. Each embodiment focuses on describing the differences from other embodiments. In particular, the embodiments for evaluating the intelligent query method for the cam configuration of a circular knitting machine, the electronic device, the electronic device, and the readable storage medium are basically similar to the embodiments of the intelligent query method for the cam configuration of a circular knitting machine described above, and are therefore described simply. Relevant parts can be referred to in the descriptions of the embodiments of the intelligent query method for the cam configuration of a circular knitting machine described above.

Claims

1. A method for intelligently querying the triangle configuration of a circular knitting machine, characterized in that, include: Collect multi-source query data including triangle model parameters, knitting process requirements, and fabric characteristics, and configure priority and adaptation weight parameters. The data preprocessing module completes the standardization transformation of heterogeneous data to form a query dataset with a unified format. The query dataset is parsed to extract core attributes, including parameter dimensions and data matching degree. The data retrieval priority is optimized by sorting the parameters according to their importance, and an optimized query dataset with both parameter completeness and retrieval adaptability is generated. By combining the complexity of query requirements with the computing power of the database, the query dataset is processed to optimize the query strategy, the search scope and the number of matching iterations are determined simultaneously, and query batches containing key parameters and matching tags are generated. For data of the same type and with consistent weighting, a parallel retrieval mechanism is used to improve import efficiency. Based on a multi-dimensional architecture of parameter indexing engine, process association model, and fit scoring module, this system establishes parameter mappings through association rule learning. It dynamically adapts weights based on knitting machine type and production scenario, optimizes the query algorithm using weighted matching logic, and simultaneously verifies fit accuracy and response efficiency in real time. In case of anomalies, the system dynamically adjusts the retrieval strategy. This includes establishing a multi-dimensional parameter mapping of triangular parameters—process—fabric—based on knitting machine type and production scenario requirements, determining the dynamic fit weight allocation logic, and ranking the results based on parameter importance. The search strategy requires optimizing the query algorithm based on weighted matching logic, clarifying the execution logic of prioritizing the comparison of core parameters and gradient matching of secondary parameters; adopting a real-time verification mechanism with two indicators, adaptation accuracy and response efficiency, setting a qualified threshold for accuracy and an upper limit for response efficiency, and continuously monitoring changes in indicators during the query process; triggering a parameter weight recalibration process for cases where adaptation accuracy fails to meet the standard; initiating a retrieval strategy downgrade adjustment for response efficiency timeout information; and conducting secondary verification of the adjusted query algorithm and retrieval strategy, synchronously recording parameter adjustment logs and strategy optimization trajectories to form a stable query execution plan that balances matching accuracy and execution efficiency. After completing the query matching, the accuracy of the results is optimized based on the model-specific rules and dynamic adaptation algorithm, and duplicate configuration schemes within the similarity threshold are eliminated through a redundancy filtering mechanism.

2. The method as described in claim 1, characterized in that, The query dataset is parsed to extract core attributes, including parameter dimensions and data matching degree. Data retrieval priorities are optimized by sorting parameters by importance, generating an optimized query dataset that combines parameter completeness and retrieval suitability, including: To address the characteristics of query datasets containing multi-source heterogeneous parameters and differences in core attribute dimensions, a full-process processing mechanism is established, encompassing data parsing, core attribute extraction, parameter importance quantification, and retrieval priority ranking, to achieve data regularization. Then, through a hierarchical optimization strategy of assigning weights to key parameters and filtering secondary parameters, and the step-by-step execution of importance scoring and priority ranking, a basic set of query data retrieval data covering all elements from attributes to weights to ranking and retrieval adaptation is formed. To meet the core requirements of parameter accuracy and retrieval efficiency for the triangle configuration query of circular knitting machines, a query data priority optimization mechanism is constructed. A parameter importance assessment module is added to the mechanism to accurately sort out and refine the retrieval needs and priority ranking criteria for different types of query data. We construct parameter attribute-retrieval priority mapping rules to transform the core feature requirements of query data into executable sorting logic. At the same time, we introduce a parameter matching degree qualification threshold to standardize and verify the sorting results. Combined with an abnormal data secondary screening mechanism, we adjust the sorting of those that do not meet the standard, and generate an optimized query dataset that has both parameter completeness and retrieval adaptability.

3. The method as described in claim 2, characterized in that, By considering the complexity of query requirements and the computing resources of the database, the query dataset is processed to optimize the search strategy, and the search scope and number of matching iterations are determined simultaneously. This generates query batches containing key parameters and appropriate tags, including: Analyze the query requirement complexity of the optimized query dataset, identify the core indicators such as process adaptation accuracy threshold, parameter matching constraints, and fabric characteristic adaptation difficulty, and complete the classification and labeling according to complexity level. Quantitatively evaluate the database computing resources, obtain key performance parameters such as data processing throughput, retrieval response latency, and concurrent processing capability, and establish a computing resource supply and demand matching evaluation table; Based on the classification results of demand complexity and the evaluation data of computing resources, the adaptability retrieval strategy is determined through negotiation using the supply and demand balance algorithm. Among them, high-complexity demand is matched with a precise retrieval strategy, and medium- and low-complexity demand is matched with a fast retrieval strategy. By combining the parameter size, parameter correlation and matching difficulty of the optimized query dataset, and adhering to the principle of full coverage of core parameters and selective filtering of secondary parameters as needed, the scope and boundaries of the retrieved data are clearly defined. Based on the selected retrieval strategy and data range, and combined with the upper limit of computing power, the number of target matching iterations is determined through iterative efficiency simulation calculations. Extract key parameters and adaptation tags from the optimized query dataset, integrate retrieval execution rules, and generate structured query batches that can be directly used for retrieval.

4. The method as described in claim 1, characterized in that, After completing the query matching, the accuracy of the results is optimized based on model-specific rules and dynamic adaptation algorithms. A redundancy filtering mechanism is used to remove duplicate configuration schemes within the similarity threshold, including: The system performs rule parsing and standardization on the model compatibility requirements, parameter matching completeness, and process compatibility consistency in the query matching results to generate model-specific verification conditions, core parameter matching thresholds, and process compatibility qualification standards, thus forming the basic information for optimizing the accuracy of the results. The parameter adjustment logic, result correction range, and accuracy improvement target of the dynamic adaptation algorithm are constrained, defined, and quantified to generate adaptation parameter optimization coefficients, result correction boundaries, and accuracy achievement thresholds, thus forming result optimization constraint information. The criteria for judging redundant configuration schemes, the dimensions of similarity calculation, and the filtering execution logic are refined and parameterized to generate duplicate scheme judgment conditions, multi-dimensional similarity calculation benchmarks, and redundancy filtering execution thresholds, thus forming redundancy removal execution information. The system integrates basic information for result accuracy optimization, result optimization constraint information, and redundant removal execution information. It performs full-process collaborative processing of query result accuracy verification, dynamic adaptation optimization, and redundant scheme removal to generate the final query result of triangular configuration for knitting circular knitting machines that combines accuracy and uniqueness.

5. A smart query system for the triangle configuration of a circular knitting machine, characterized in that, The system is configured to execute the intelligent query method for the triangular configuration of a circular knitting machine as described in any one of claims 1 to 4 by executing executable instructions.

6. An electronic device, characterized in that, include: First processor; and memory for storing executable instructions of the first processor; The first processor is configured to execute the intelligent query method for the triangular configuration of a circular knitting machine according to any one of claims 1 to 4 by executing the executable instructions.

7. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the second processor, it implements the intelligent query method for the triangular configuration of the circular knitting machine as described in any one of claims 1 to 4.

Images