A motor selection method and system

Abstract

The application relates to the technical field of motor selection, in particular to a motor selection method and system, which comprises the following steps: constructing a distributed data query library, introducing a standardized feature vector into the distributed data query library to match a candidate motor model set; obtaining comprehensive performance scores of each candidate motor model, triggering a model elimination mechanism based on a preset score value threshold, performing performance parameter sorting on the retained candidate motor models to generate an optimal motor model set; obtaining comprehensive energy efficiency scores of each optimal motor model and establishing a Pareto frontier solution set for secondary screening to obtain a recommended motor model set; and dividing the recommended motor model into multiple motor selection levels and taking the motor selection levels as the final motor selection result output. The application improves the scientificity, accuracy and practicability of motor selection through an intelligent motor selection process, and provides high-quality selection decision support for users.

Description

A method and system for selecting motors Technical Field

[0001] This invention relates to the field of motor selection technology, and in particular to a motor selection method and system. Background Technology

[0002] In modern logistics and warehousing systems, the selection of linear RGV motors and heavy-duty stacker crane motors is a key technical aspect to ensure efficient system operation. With the increasing demands for automation and intelligence in the logistics industry, traditional motor selection methods are no longer sufficient to meet the performance requirements under complex operating conditions. As the core drive unit of rail-mounted transport equipment, the selection of linear RGV motors requires careful consideration of parameters such as acceleration, positioning accuracy, load adaptability, and energy efficiency. Meanwhile, motors for heavy-duty stacker cranes need to meet higher standards in terms of load-bearing capacity, operational stability, impact resistance, and long-term reliability. However, existing motor selection methods often rely on empirical formulas or single performance indicators, lacking the ability to collaboratively optimize multi-dimensional parameters, resulting in selection results that often fail to fully meet actual needs. Especially when facing diverse operating conditions, traditional methods struggle to dynamically adjust weight allocation and accurately match performance priorities under different scenarios. Furthermore, existing technologies do not adequately consider the full life-cycle cost of motors and lack the ability to predict energy efficiency degradation trends, potentially leading to excessively high maintenance costs in the later stages. In terms of data processing, traditional methods often employ a centralized query model, which struggles to meet the rapid matching needs of massive amounts of motor model data and lacks the ability to adaptively correct for abnormal data. At the decision support level, existing technologies typically output a single recommendation result, failing to establish a hierarchical set of selection solutions and thus unable to provide users with multi-dimensional decision-making references. Therefore, there is an urgent need for a motor selection method that can integrate multi-source data, support dynamic optimization, and provide hierarchical decision-making to improve the scientific rigor, accuracy, and practicality of selecting linear RGV motors and heavy-duty stacker crane motors. Summary of the Invention

[0003] This invention overcomes the shortcomings of the prior art and provides a method and system for selecting motors.

[0004] To achieve the above objectives, the technical solution adopted by the present invention is as follows:

[0005] The first aspect of this invention discloses a method for selecting a motor, comprising the following steps:

[0006] Obtain the selection benchmark parameters input by the user, convert the selection benchmark parameters into a benchmark measurement system, and obtain a standardized feature vector;

[0007] Construct a distributed data query library, and import the standardized feature vector into the distributed data query library to match the candidate motor model set;

[0008] Obtain the comprehensive performance score of each candidate motor model, trigger the model elimination mechanism based on the preset score threshold, and sort the remaining candidate motor models by performance parameters to generate a set of preferred motor models;

[0009] Obtain the comprehensive energy efficiency score of each preferred motor model and establish a Pareto front solution set for secondary screening to obtain a recommended set of motor models;

[0010] The recommended motor models are divided into multiple motor selection levels, and the motor selection levels are output as the final motor selection result.

[0011] Preferably, the selection benchmark parameters are converted into a benchmark metrology system to obtain a standardized feature vector, specifically as follows:

[0012] Extract the numerical values ​​and unit identifiers of each input selection benchmark parameter, and dynamically construct a parameter-unit mapping table;

[0013] The unit identifier of each parameter in the mapping table is identified. If the unit is imperial or non-standard metric, the corresponding selection benchmark parameter is converted into an equivalent value under the benchmark unit.

[0014] Based on the converted selection benchmark parameters, each parameter is range-mapped according to the preset threshold range of motor performance parameters;

[0015] The normalized numerical parameters and the encoded discrete parameters are concatenated by the feature fusion layer according to their weights and priorities to form a multi-dimensional standardized feature vector. The order of the vector dimensions is dynamically adjusted according to the key influencing factors of the motor selection decision tree.

[0016] The standardized feature vectors are subjected to integrity checks, and missing parameters are filled by interpolation under adjacent operating conditions or fitting historical data, and the output is a standardized feature vector set that meets the input requirements of the motor performance matching model.

[0017] Preferably, a distributed data query library is constructed, and the standardized feature vector is imported into the distributed data query library to match the candidate motor model set, specifically as follows:

[0018] The system acquires the baseline parameter data streams of motors from various manufacturers, converts them into standardized parameters with unified dimensions using a pre-defined motor industry standard terminology mapping table, and constructs a standardized parameter table.

[0019] The standardized parameter table is matrixed to form a multidimensional feature matrix, where each row of the matrix represents a motor model and each column corresponds to a standardized parameter feature; a data query library is constructed by importing the multidimensional feature matrix into the data query library to obtain a distributed data query library;

[0020] The standardized feature vectors are expanded in dimension to match the column dimensions of the multidimensional feature matrix, thus forming a query vector matrix;

[0021] The query vector is compared with each row in the feature matrix, and the Euclidean distance between them is calculated in each feature dimension. If the Euclidean distance of a certain row is less than the preset distance threshold, the motor model of the corresponding row is marked as a candidate model.

[0022] The candidate model set is prioritized and sorted from smallest to largest according to Euclidean distance to generate the final candidate motor model set;

[0023] Throughout the process, if the set of candidate motor models is empty, the preset distance threshold will be automatically adjusted or the user will be prompted to re-enter the selection benchmark parameters.

[0024] Preferably, the comprehensive performance score of each candidate motor model is obtained, and a model elimination mechanism is triggered based on a preset score threshold. The remaining candidate motor models are then sorted by performance parameters to generate a set of preferred motor models. Specifically:

[0025] Key performance parameters of each candidate motor model are extracted from the set of candidate motor models, including rated power curves, efficiency distribution maps and temperature rise characteristic data, and a three-dimensional performance matrix is ​​constructed as a benchmark data source for multi-dimensional parameter comparison.

[0026] Based on the application scenario characteristics input by the user, the torque response coefficient, energy efficiency priority factor and environmental adaptation weight are automatically allocated to form a dynamic weight allocation rule. If the scenario characteristics are not clear, the default weight allocation mode is activated and the log is recorded.

[0027] Based on dynamic weighting rules, the parameters in the three-dimensional performance matrix are weighted and calculated to generate a comprehensive performance score for each candidate motor model; the comprehensive performance score of each candidate motor model is then compared with a preset score threshold.

[0028] If the overall performance score of a candidate motor model is less than the preset score threshold, the model elimination mechanism will be triggered to remove the candidate motor model from the candidate motor model set.

[0029] The retained candidate models are sorted from highest to lowest comprehensive performance score to generate a set of preferred motor models;

[0030] If the scores are the same, a second sorting will be performed based on the energy efficiency priority factor to ensure the uniqueness of the sorting results, and the preferred sequence will be used as the input data for the next stage of load characteristic matching verification.

[0031] Preferably, the comprehensive energy efficiency score of each preferred motor model is obtained, and a Pareto front solution set is established for secondary screening to obtain a recommended set of motor models, specifically:

[0032] Obtain rated operating condition data for each preferred motor model, and extract efficiency distribution characteristics of each preferred motor model under different load points based on the rated operating condition data;

[0033] Based on the energy efficiency priority parameters input by the user, efficiency weight coefficients are dynamically allocated, and combined with the efficiency distribution characteristics of each preferred motor model under different load points, the comprehensive energy efficiency score of each preferred motor model is calculated; if the comprehensive energy efficiency score of a certain preferred motor model is lower than the preset score threshold, the model downgrade mechanism is triggered, and the model is removed from the preferred queue.

[0034] A multi-objective particle swarm optimization algorithm is adopted, with comprehensive energy efficiency score, power density and temperature rise characteristics as optimization objectives, to establish Pareto front solution sets for each preferred motor model; if the number of Pareto front solution sets exceeds the preset upper limit, a fuzzy comprehensive evaluation method is introduced to perform secondary screening of the solution sets until the number of Pareto front solution sets is lower than the preset upper limit.

[0035] Subsequently, the embedded simulation engine was invoked to perform virtual operating condition loading tests on each preferred motor model based on the Pareto front solution set of each preferred motor model, and the simulated energy efficiency degradation curve of each preferred motor model was obtained.

[0036] The simulated energy efficiency decay curves of each preferred motor model are compared with the preset energy efficiency decay curves to obtain the degree of overlap between the simulated energy efficiency decay curves and the preset energy efficiency decay curves of each preferred motor model.

[0037] If the overlap between the simulated energy efficiency degradation curve of a preferred motor model and the preset energy efficiency degradation curve is not greater than the preset overlap, the model downgrade mechanism is triggered, and the model is removed from the preferred queue.

[0038] Obtain the remaining preferred motor models from the preferred motor model set to obtain the recommended motor model set.

[0039] Preferably, a multi-objective particle swarm optimization algorithm is used, with comprehensive energy efficiency score, power density, and temperature rise characteristics as optimization objectives, to establish the Pareto front solution set for each preferred motor model, specifically as follows:

[0040] Initialize the particle swarm by using the comprehensive energy efficiency score, power density, and temperature rise characteristics of each preferred motor model as the three-dimensional target vector of the particles, and randomly assign initial positions and velocities.

[0041] Calculate the fitness value of each particle and construct an initial Pareto solution set; if the number of solutions is zero, adjust the target weight coefficient and recalculate the fitness value.

[0042] The particle position update algorithm is executed to dynamically adjust the particle trajectory based on the individual optimal solution and the global optimal solution. If a particle is detected to be trapped in a local optimum, a mutation operation is introduced to randomly perturb the particle position in order to escape the local extremum.

[0043] The updated particles are classified using a non-dominated sorting method to select a new Pareto solution set; if the number of solutions exceeds a preset limit, crowding calculation is initiated, and solutions with uniform distribution are retained first.

[0044] Meanwhile, a solution set convergence judgment mechanism is established. If the number of consecutive iterations exceeds the threshold and the solution set change rate is lower than the preset value, the algorithm is judged to have converged; if it has not converged, iterative optimization continues.

[0045] Finally, the final Pareto front solution set is output, and the corresponding motor model and key parameters of each solution are labeled.

[0046] Preferably, if the number of solutions to the Pareto front exceeds a preset upper limit, a fuzzy comprehensive evaluation method is introduced to perform a secondary screening of the solutions until the number of solutions to the Pareto front falls below the preset upper limit, specifically as follows:

[0047] Extract the comprehensive energy efficiency score, power density, and temperature rise characteristics of each motor model from the Pareto front solution set, and construct a multi-dimensional evaluation matrix;

[0048] The weight coefficients of each evaluation indicator are dynamically allocated based on the energy efficiency priority parameters input by the user.

[0049] Establish a fuzzy membership function to normalize each evaluation index and eliminate dimensional differences; if uneven distribution of the normalized data is detected, adjust the membership function parameters.

[0050] The weighted average method is used to calculate the comprehensive fuzzy evaluation value of each motor model, and the solution set is sorted according to the comprehensive fuzzy evaluation value;

[0051] Set an upper limit threshold for the number of solution sets, and sequentially remove the motor models with the lowest comprehensive fuzzy evaluation values ​​until the number of motor models after removal is less than the upper limit threshold for the number of solution sets; if the number of solution sets after removal still exceeds the upper limit, gradually increase the removal criteria until the quantity requirement is met.

[0052] Finally, the filtered Pareto front solution set is output, and the comprehensive fuzzy evaluation value and key parameters of each model are labeled.

[0053] Preferably, the recommended motor models are divided into multiple motor selection levels, specifically:

[0054] Extract the comprehensive energy efficiency score, cost-effectiveness index and compatibility parameters of all models in the recommended motor model set, and construct a three-dimensional classification feature matrix;

[0055] Based on the core performance thresholds set by the user, a set of models with comprehensive energy efficiency scores higher than the benchmark and compatibility standards are selected and marked as the primary recommended candidate set; if the candidate set is empty, the threshold range is automatically widened and the adjustment parameters are recorded.

[0056] For the remaining models, perform an economic analysis, calculate the life cycle cost-benefit ratio, and screen out models with a cost-benefit ratio lower than the industry average and compatibility parameters within the acceptable range, classifying them as economic substitutes; if missing cost data is detected, call the historical procurement database to fill in the substitute values;

[0057] For models not selected as the primary recommended or economical alternatives, extended compatibility verification will be conducted, and models that pass the extended compatibility verification will be designated as extended compatible models.

[0058] The selection criteria parameters include gross weight under full load, speed under full load, acceleration under full load, diameter of traveling wheels / drum, energy efficiency level requirements, three-dimensional size limitations of installation space, energy efficiency priority parameters, environmental temperature and humidity gradient parameters, protection level code, expected life index, and application scenario characteristics.

[0059] The second aspect of the present invention discloses a motor selection system, the motor selection system including a memory and a processor, the memory storing a motor selection method program, and when the motor selection method program is executed by the processor, the steps of any of the motor selection methods described in the present invention are implemented.

[0060] This invention addresses the technical deficiencies in the prior art and offers the following advantages: It acquires user-inputted selection benchmark parameters and converts them into a benchmark measurement system to obtain standardized feature vectors; it constructs a distributed data query library and imports the standardized feature vectors into the library to match a set of candidate motor models; it obtains the comprehensive performance score of each candidate motor model, triggers a model elimination mechanism based on a preset score threshold, and sorts the remaining candidate motor models by performance parameters to generate a set of preferred motor models; it obtains the comprehensive energy efficiency score of each preferred motor model and establishes a Pareto front solution set for secondary screening to obtain a set of recommended motor models; it divides the recommended motor models into multiple motor selection levels and outputs these levels as the final motor selection result. This invention, through an intelligent motor selection process, improves the scientific rigor, accuracy, and practicality of motor selection, providing users with high-quality selection decision support. Attached Figure Description

[0061] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other embodiments can be obtained from these drawings without creative effort.

[0062] Figure 1 is a flowchart of an overall method for motor selection;

[0063] Figure 2 is a partial flowchart of a motor selection method;

[0064] Figure 3 is a system block diagram of a motor selection system;

[0065] Figure 4 is a schematic diagram of a comparative example of motor energy efficiency degradation curves;

[0066] Figure 5 shows the trajectory of a single particle.

[0067] Figure 6 shows the particle swarm distribution and Pareto front plot. Detailed Implementation

[0068] To better understand the above-mentioned objectives, features, and advantages of the present invention, the present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments. It should be noted that, unless otherwise specified, the embodiments and features described in these embodiments can be combined with each other.

[0069] Many specific details are set forth in the following description in order to provide a full understanding of the invention. However, the invention may also be practiced in other ways different from those described herein, and therefore the scope of protection of the invention is not limited to the specific embodiments disclosed below.

[0070] As shown in Figure 1, the first aspect of this invention discloses a method for selecting a motor, comprising the following steps:

[0071] S102. Obtain the selection benchmark parameters input by the user, convert the selection benchmark parameters into a benchmark measurement system, and obtain a standardized feature vector;

[0072] S104. Construct a distributed data query library, and import the standardized feature vector into the distributed data query library to match the candidate motor model set;

[0073] S106. Obtain the comprehensive performance score of each candidate motor model, trigger the model elimination mechanism based on the preset score threshold, and sort the remaining candidate motor models by performance parameters to generate a set of preferred motor models.

[0074] S108. Obtain the comprehensive energy efficiency score of each preferred motor model and establish a Pareto front solution set for secondary screening to obtain a set of recommended motor models;

[0075] S110. Divide the recommended motor models into multiple motor selection levels, and output the motor selection levels as the final motor selection result.

[0076] It should be noted that by converting the user-input selection benchmark parameters into a benchmark measurement system and generating standardized feature vectors, the consistency and comparability of the data are ensured, providing a scientific data foundation for subsequent matching and screening. Constructing a distributed data query library and importing standardized feature vectors for matching improves data processing efficiency and query speed, ensuring the comprehensiveness and accuracy of the candidate motor model set. Next, a model elimination mechanism is triggered based on comprehensive performance scores, and the remaining candidate models are sorted by performance parameters to ensure the high quality and suitability of the preferred motor model set. Then, the selection results are further optimized through comprehensive energy efficiency score calculation and Pareto front solution set construction, ensuring the comprehensiveness and optimality of the solution set; if the number of solution sets exceeds a preset limit, a secondary screening mechanism is introduced to improve the accuracy of the selection results. Finally, the recommended motor models are divided into multiple selection levels, providing users with hierarchical and multi-dimensional selection decision support, ensuring the practicality and flexibility of the selection results. In summary, this invention can effectively improve the accuracy and efficiency of motor selection, providing reliable technical support for motor selection decisions.

[0077] Preferably, the selection benchmark parameters are converted into a benchmark metrology system to obtain a standardized feature vector, as shown in Figure 2, specifically:

[0078] S202. Extract the values ​​and unit identifiers of each input selection benchmark parameter, and dynamically construct a parameter-unit mapping table;

[0079] S204. Identify the unit identifier of each parameter in the mapping table. If the unit is imperial or non-standard metric, convert the corresponding selection benchmark parameter into the equivalent value under the benchmark unit.

[0080] It should be noted that, firstly, the system dynamically parses the unit identifier (such as the speed unit "mph" or the length unit "inch") of each parameter in the parameter-unit mapping table. If it is identified as an imperial unit (such as mph, inch) or a non-standard metric unit (such as the non-ISO standard "kgf"), it calls the preset unit conversion rule library (such as 1 mph = 0.44704 m / s, 1 inch = 25.4 mm) to perform the base unit conversion. If the unit is missing (such as the user not specifying a speed unit), it matches the default unit based on the parameter type (such as the default speed unit being "m / s"). After the conversion, the equivalent value is updated in the mapping table, and the original unit and conversion log are recorded to ensure data traceability. For example, "6 inches" is converted to 152.4 mm, and the conversion record from "inches to mm" is retained. The preset unit conversion rule library is a pre-defined database that stores conversion relationships between various units and benchmark units. It includes conversion coefficients for imperial, metric, and other industry-specific units. It is used to uniformly convert parameters of different unit systems into standard benchmark units (such as uniform length to millimeters and speed to meters per second) to ensure data consistency and comparability.

[0081] S206. For the converted selection benchmark parameters, perform interval mapping on each parameter according to the preset threshold range of motor performance parameters;

[0082] S208. The normalized numerical parameters and the encoded discrete parameters are concatenated by the feature fusion layer according to their weights and priorities to form a multi-dimensional standardized feature vector. The order of the vector dimensions is dynamically adjusted according to the key influencing factors of the motor selection decision tree.

[0083] It should be noted that, based on the key influencing factors of the motor selection decision tree, weight values ​​are assigned to each normalized numerical parameter and coded discrete parameter. These weight values ​​are determined according to the parameter's influence on the selection result. Next, the parameters are sorted from highest to lowest weight, with higher-weight parameters appearing at the beginning of the feature vector and lower-weight parameters appearing later. If weights are the same, a secondary sort is performed based on the parameter's data type (continuous parameters take precedence over discrete parameters). Then, the sorted parameters are concatenated in order to form a multi-dimensional standardized feature vector, ensuring that the vector dimension order is consistent with the key influencing factors of the motor selection decision tree.

[0084] S210. Perform integrity checks on the standardized feature vectors, and fill in the missing parameters by interpolation under adjacent operating conditions or fitting historical data, and output a standardized feature vector set that meets the input requirements of the motor performance matching model.

[0085] In a preferred embodiment of the present invention, in the motor selection scenario of a logistics warehousing AGV, the user-input selection benchmark parameters include the full load weight (1200kg, with "kg" as the default if the unit is missing), the full load speed (2.5mph, imperial units), the wheel diameter (6 inches), the energy efficiency rating (IE3), and the installation space limitation (length × width × height = 400 × 300 × 200mm). After the data parsing module extracts the parameter values ​​and unit identifiers, it dynamically constructs a parameter-unit mapping table (e.g., speed is mapped as "mph → m / s", which is converted to an equivalent value of 1.1176m / s using a formula; the wheel diameter of 6 inches is converted to 152.4mm). Subsequently, the parameters are range-mapped based on a preset motor performance threshold range (e.g., rated power threshold 0.5-5kW): the full load weight of 1200kg is normalized to 0.6 through linear scaling (threshold 0-2000kg), and the speed of 1.1176m / s (threshold 0-3m / s) is mapped to 0.3725. The discrete parameter energy efficiency level IE3 is generated into a binary vector [0,1,0] through one-hot encoding. The feature fusion layer concatenates the normalized values ​​with the encoded vector according to weight priority (rated power weight 0.3 > efficiency 0.25 > size 0.2) to form a standardized feature vector [0.6, 0.3725, 0.76 (wheel diameter normalized value), 0,1,0], where the wheel diameter is shifted forward due to the "installation compatibility" factor in the decision tree. If a parameter is missing (e.g., acceleration is not input), interpolation based on adjacent operating conditions is used (the average acceleration in historical AGV data is 0.8 m / s², which is padded to 0.8). Finally, a standardized feature vector set is output for subsequent distributed query library matching of candidate motor models.

[0086] In summary, the systematic data processing workflow improves the standardization and usability of selection benchmark parameters, providing an accurate and reliable data foundation for motor selection decisions, thereby enhancing the efficiency and accuracy of motor selection.

[0087] Preferably, a distributed data query library is constructed, and the standardized feature vector is imported into the distributed data query library to match the candidate motor model set, specifically as follows:

[0088] The system acquires the baseline parameter data streams of motors from various manufacturers, converts them into standardized parameters with unified dimensions using a pre-defined motor industry standard terminology mapping table, and constructs a standardized parameter table.

[0089] Among them, the preset motor industry standard terminology mapping table is a pre-defined reference table used to unify the naming and dimensions of motor parameters from different manufacturers. It maps the heterogeneous parameter terms (such as "output power" and "rated power") and units (such as "HP" and "kW") of various manufacturers to standard terms and benchmark units to ensure data comparability and calculation consistency.

[0090] It should be noted that the data acquisition module extracts benchmark parameter data streams from motor technical documents provided by various manufacturers, including key parameters such as rated power, efficiency, and speed. Next, a pre-defined motor industry standard terminology mapping table is loaded, matching the parameter terms used by each manufacturer with the standard terms to unify parameter naming and unit identification. If an undefined term is found, a terminology expansion mechanism is activated, intelligently matching and updating the mapping table based on contextual semantics. Then, the uniformly named parameters undergo unit conversion, converting imperial or non-standard metric units to benchmark units to ensure all parameters have a unified dimension.

[0091] The standardized parameter table is matrixed to form a multidimensional feature matrix, where each row of the matrix represents a motor model and each column corresponds to a standardized parameter feature; a data query library is constructed by importing the multidimensional feature matrix into the data query library to obtain a distributed data query library;

[0092] The standardized feature vectors are expanded in dimension to match the column dimensions of the multidimensional feature matrix, thus forming a query vector matrix;

[0093] It should be noted that, firstly, the column dimensions of the multidimensional feature matrix are defined (e.g., the standardized parameter order such as rated power, speed, and size). The parameters in the standardized feature vector are then arranged in the same dimensional order. If a parameter corresponding to a matrix column is missing from the vector (e.g., historical data does not record the protection level), zero values ​​are used for padding or missing terms are imputed using the mean of parameters from similar motors. Subsequently, the feature vector length is adjusted to match the number of columns in the multidimensional feature matrix, ensuring a one-to-one correspondence between each dimension, ultimately forming a query vector matrix perfectly aligned with the matrix column structure. For example, if the matrix contains 6 columns of parameters, and the original feature vector only has 5 columns, default values ​​of 0 are added to the end to complete the expansion.

[0094] The query vector is compared with each row in the feature matrix, and the Euclidean distance between them is calculated in each feature dimension. If the Euclidean distance of a certain row is less than the preset distance threshold, the motor model of the corresponding row is marked as a candidate model.

[0095] The formula for calculating Euclidean distance is as follows:

[0096] ;

[0097] In the formula, Represents the query vector With the first in the characteristic matrix Euclidean distance of the rows; Indicates the query vector at the th The value of the dimension; Represents the th element in the characteristic matrix line, number The value of the column.

[0098] The candidate model set is prioritized and sorted from smallest to largest according to Euclidean distance to generate the final candidate motor model set;

[0099] Throughout the process, if the set of candidate motor models is empty, the preset distance threshold will be automatically adjusted or the user will be prompted to re-enter the selection benchmark parameters.

[0100] In the scenario of motor selection for AGVs in logistics warehousing, when constructing a distributed data query library based on the standardized feature vectors generated in previous embodiments (such as [0.6, 0.3725, 0.76, 0,1,0]), the first step is to obtain the motor parameters of manufacturer A (such as "output power: 3.7HP", "efficiency: 92%", "speed: 2800rpm") and the parameters of manufacturer B (such as "rated power: 2.2kW", "efficiency: IE4", "speed: 1500r / min") through the data acquisition module. Using a preset motor industry standard terminology mapping table, "output power" is unified to "rated power" (unit converted to kW, such as 3.7HP→2.76kW), "efficiency: IE4" is mapped to "energy efficiency level code 4", and the speed unit is unified to "r / min". Subsequently, the standardized parameter table matrix is ​​transformed into a 3x6 multidimensional feature matrix (e.g., the model data for manufacturer C is [0.65 (normalized rated power), 0.35 (normalized speed), 0.72 (installation size score), 0,1,0 (energy efficiency code)]). After expanding the standardized feature vector into a 6-dimensional query vector matrix, the Euclidean distance is calculated row by row: for example, the distance between query vector Q and manufacturer C model is approximately 0.18. If the preset distance threshold is 0.2, then this model is marked as a candidate. Finally, a candidate set is generated in ascending order of distance (e.g., manufacturer C model distance 0.18 > manufacturer B model distance 0.25). If the initial candidates are empty, the threshold is automatically adjusted to 0.3 to include more models, ensuring that the matching results are suitable for the low-speed, high-torque requirements of AGVs.

[0101] In summary, this step, through a systematic data processing and matching process, can improve the efficiency, accuracy, and adaptability of motor model matching, providing reliable technical support for motor selection decisions.

[0102] Preferably, the comprehensive performance score of each candidate motor model is obtained, and a model elimination mechanism is triggered based on a preset score threshold. The remaining candidate motor models are then sorted by performance parameters to generate a set of preferred motor models. Specifically:

[0103] Key performance parameters of each candidate motor model are extracted from the set of candidate motor models, including rated power curves, efficiency distribution maps and temperature rise characteristic data, and a three-dimensional performance matrix is ​​constructed as a benchmark data source for multi-dimensional parameter comparison.

[0104] It should be noted that rated power curve data, including the output power values ​​of the motor at different load points, are extracted from the technical documents or test reports provided by various motor manufacturers. If data is missing, it is supplemented by interpolation based on historical data of similar models. Next, the efficiency values ​​of the motor at different loads and speeds are extracted from the efficiency test reports to generate efficiency distribution maps. If the test conditions are incomplete, simulation models are used to supplement the data. Then, temperature rise characteristic data of the motor under different operating conditions, including steady-state temperature and temperature rise rate, are obtained through temperature rise experiments or thermal simulations. If experimental data is insufficient, it is calculated based on thermodynamic models. Finally, the extracted data is verified and standardized to ensure that the data format is uniform and there are no outliers, and a complete performance parameter dataset is constructed as the basis for subsequent analysis.

[0105] Based on the application scenario characteristics input by the user, the torque response coefficient, energy efficiency priority factor and environmental adaptation weight are automatically allocated to form a dynamic weight allocation rule. If the scenario characteristics are not clear, the default weight allocation mode is activated and the log is recorded.

[0106] Among them, "application scenario characteristics" refers to key information such as the usage environment, working conditions and performance requirements specifically described by users during the motor selection process, including but not limited to load type (such as constant load, periodic load), operating environment (such as high temperature, humidity, dust), energy efficiency requirements (such as high energy efficiency priority, cost priority), operating frequency (such as continuous operation, intermittent operation) and special requirements (such as low noise, high precision).

[0107] It should be noted that the process involves parsing the application scenario features input by the user and extracting key parameters such as load type, operating environment, and energy efficiency requirements. If unclear scenario features or missing parameters are detected, a default scenario template is used to complete the feature set and the process is logged. Next, based on a pre-defined scenario feature-weight mapping table, the initial values ​​of the torque response coefficient, energy efficiency priority factor, and environmental adaptation weight are dynamically calculated according to the extracted key parameters. If the current scenario is not defined in the mapping table, a weight inference mechanism is initiated, estimating weights based on historical data from similar scenarios. Then, the initial weight values ​​are normalized to ensure that the sum of the three is 1. Subsequently, the final weight values ​​are written to the dynamic weight allocation rule base, and a weight allocation report is generated for user confirmation. If the user requests modifications, the weight allocation process is re-executed. Finally, the complete weight allocation process and result logs are recorded as a reference for subsequent weight rule optimization.

[0108] Based on dynamic weighting rules, the parameters in the three-dimensional performance matrix are weighted and calculated to generate a comprehensive performance score for each candidate motor model; the comprehensive performance score of each candidate motor model is then compared with a preset score threshold.

[0109] It should be noted that, firstly, the torque response coefficient, energy efficiency priority factor, and environmental adaptation weight value for the current scenario are obtained from the dynamic weight rule base. Next, the rated power curve, efficiency distribution map, and temperature rise characteristic data of each candidate motor model are extracted from the three-dimensional performance matrix, and then multiplied by their corresponding weight values ​​to calculate the weighted score for each parameter. Finally, the weighted scores of each parameter are summed to generate a comprehensive performance score for each candidate motor model, calculated using the following formula:

[0110] ;

[0111] In the formula, The overall performance score for motor model M; The torque response coefficient represents the weight of rated power in relation to scenario requirements. This is the energy efficiency priority factor, representing the weight of efficiency in relation to scenario requirements; The environmental adaptation weight represents the weight of temperature rise characteristics on scenario requirements; This is the normalized mean of the rated power; This is the normalized mean of efficiency; This is the normalized inverse mean of the temperature rise characteristics.

[0112] If the overall performance score of a candidate motor model is less than the preset score threshold, the model elimination mechanism will be triggered to remove the candidate motor model from the candidate motor model set.

[0113] The retained candidate models are sorted from highest to lowest comprehensive performance score to generate a set of preferred motor models;

[0114] If the scores are the same, a second sorting will be performed based on the energy efficiency priority factor to ensure the uniqueness of the sorting results, and the preferred sequence will be used as the input data for the next stage of load characteristic matching verification.

[0115] Similarly, in the motor selection scenario for AGVs in logistics warehousing, based on the candidate model set matched in the previous embodiment (such as model M1 from manufacturer C and model M2 from manufacturer B), the rated power curve (e.g., 2.1kW output at 50% load, 3.0kW at 100% load), efficiency distribution map (92% efficiency at 50% load, 88% efficiency at 100% load), and temperature rise characteristic data (steady-state temperature of 65°C after 2 hours of continuous operation) of M1 are first extracted from the manufacturer's technical documents to construct a three-dimensional performance matrix. According to the user-input scenario characteristics of "high-temperature warehousing environment, high energy efficiency priority", the system dynamically allocates weights: torque response coefficient α=0.4 (emphasizing load adaptability), energy efficiency priority factor β=0.5 (high efficiency requirement), and environmental adaptability weight γ=0.1 (low temperature rise tolerance). Calculate the comprehensive score for M1: The average rated power is normalized to 0.72, the average efficiency is 0.9, and the average temperature rise is 0.6. Substituting these values ​​into the comprehensive performance scoring formula, the comprehensive performance score is calculated to be 0.798. If the preset scoring threshold is 0.75, M1 meets the standard and is retained, while M2, with a score of 0.68, is eliminated. Finally, the scores are sorted (e.g., M1 > Manufacturer D model M3's 0.78). If the scores are the same, they are sorted a second time by energy efficiency factor (e.g., M3 and M4 both have a score of 0.78, but M3 has a higher energy efficiency factor), generating an optimal set for subsequent energy efficiency verification.

[0116] In summary, by constructing a three-dimensional performance matrix, dynamically allocating weights, calculating comprehensive performance scores, eliminating models, and ranking performance, efficient and accurate motor model selection and optimization are achieved, improving the efficiency, accuracy, and adaptability of motor model selection.

[0117] Preferably, the comprehensive energy efficiency score of each preferred motor model is obtained, and a Pareto front solution set is established for secondary screening to obtain a recommended set of motor models, specifically:

[0118] Obtain rated operating condition data for each preferred motor model, and extract efficiency distribution characteristics of each preferred motor model under different load points based on the rated operating condition data;

[0119] It should be noted that rated operating condition data, including key parameters such as rated voltage, rated current, rated speed, and rated load, are extracted from the technical documents or test reports provided by the motor manufacturer. Next, efficiency values ​​of the motor at different load points (e.g., 25%, 50%, 75%, 100% load) are obtained through efficiency test reports or simulation models, generating an efficiency distribution map. Then, the extracted efficiency distribution characteristic data is verified; if outliers or data deviations are found, corrections are made based on adjacent load points or historical data. Subsequently, the verified efficiency distribution characteristic data is stored according to motor model, constructing an efficiency distribution characteristic database. Finally, the efficiency distribution characteristic data of each preferred motor model is output as the basic data source for subsequent comprehensive energy efficiency score calculations.

[0120] Based on the energy efficiency priority parameters input by the user, efficiency weight coefficients are dynamically allocated, and combined with the efficiency distribution characteristics of each preferred motor model under different load points, the comprehensive energy efficiency score of each preferred motor model is calculated; if the comprehensive energy efficiency score of a certain preferred motor model is lower than the preset score threshold, the model downgrade mechanism is triggered, and the model is removed from the preferred queue.

[0121] It should be noted that the process involves parsing the user-inputted energy efficiency priority parameters and extracting key indicators such as high-load efficiency priority, average efficiency priority, or low-load efficiency priority. Next, based on a preset energy efficiency priority-weight mapping table, efficiency weight coefficients are dynamically assigned to each load point (e.g., 25%, 50%, 75%, 100% load) according to the extracted key indicators. Then, the weight coefficients are normalized to ensure that the sum of the weights for each load point is 1. Finally, the efficiency values ​​of each load point are weighted and calculated with their corresponding weight coefficients to generate a comprehensive energy efficiency score for each preferred motor model. The calculation formula is as follows:

[0122] ;

[0123] In the formula, For comprehensive energy efficiency rating; For the first The dynamic weight of each load point is allocated according to the user's energy efficiency priority (e.g., ...). , ),satisfy ; For the first Efficiency at each load point (in %), obtained through actual measurement or simulation (e.g.) ); Number of user-defined load conditions (e.g.) (corresponding to 25%, 50%, 75%, and 100%).

[0124] A multi-objective particle swarm optimization algorithm is adopted, with comprehensive energy efficiency score, power density and temperature rise characteristics as optimization objectives, to establish Pareto front solution sets for each preferred motor model; if the number of Pareto front solution sets exceeds the preset upper limit, a fuzzy comprehensive evaluation method is introduced to perform secondary screening of the solution sets until the number of Pareto front solution sets is lower than the preset upper limit.

[0125] Subsequently, the embedded simulation engine was invoked to perform virtual operating condition loading tests on each preferred motor model based on the Pareto front solution set of each preferred motor model, and the simulated energy efficiency degradation curve of each preferred motor model was obtained.

[0126] The aforementioned simulation engine refers to a software tool based on physical models and mathematical algorithms, used to simulate the operating status and performance of motors under different working conditions. Its core is to provide a scientific basis for motor selection, performance optimization and fault prediction by inputting motor parameters (such as rated power, efficiency, temperature rise Fi) and operating conditions (such as load, speed, ambient temperature, and real-time operating data, including energy efficiency, temperature rise, vibration, etc.).

[0127] It should be noted that key parameters for each preferred motor model, including rated power, efficiency distribution, and temperature rise characteristics, are extracted from the Pareto front solution set to construct a virtual operating condition input dataset. Next, the embedded simulation engine's runtime environment is configured, the motor physical model and operating condition simulation scenario are loaded, and test parameters such as test time, load fluctuation frequency, and ambient temperature are set. Then, the simulation engine is started, and virtual operating condition loading tests are performed sequentially for each preferred motor model, recording the motor's energy efficiency values ​​at different time points to generate raw data on energy efficiency changes over time. Subsequently, the raw data is smoothed to remove noise and extract the energy efficiency degradation trend, generating a simulated energy efficiency degradation curve.

[0128] The simulated energy efficiency decay curves of each preferred motor model are compared with the preset energy efficiency decay curves to obtain the degree of overlap between the simulated energy efficiency decay curves and the preset energy efficiency decay curves of each preferred motor model.

[0129] If the overlap between the simulated energy efficiency degradation curve of a preferred motor model and the preset energy efficiency degradation curve is not greater than the preset overlap, the model downgrade mechanism is triggered, and the model is removed from the preferred queue.

[0130] For example, as shown in Figure 4, the preset energy efficiency degradation curve (dashed line) serves as a standard reference curve representing the energy efficiency degradation trend under ideal conditions. As can be seen from Figure 4, the curve for motor C deviates significantly from the preset curve, thus removing it from the preferred selection queue.

[0131] Obtain the remaining preferred motor models from the preferred motor model set to obtain the recommended motor model set.

[0132] It should be noted that the comprehensive energy efficiency score is dynamically calculated based on the efficiency distribution characteristics extracted from rated operating condition data and combined with the energy efficiency priority parameters input by the user, ensuring the scientific nature and adaptability of the score results. Furthermore, a model downgrading mechanism is triggered by a preset score threshold, improving the accuracy of the selection. Secondly, a Pareto front solution set is constructed using a multi-objective particle swarm optimization algorithm, comprehensively considering the comprehensive energy efficiency score, power density, and temperature rise characteristics to ensure the comprehensiveness and optimality of the solution set. If the number of solutions exceeds a preset upper limit, a fuzzy comprehensive evaluation method is introduced for secondary screening to further optimize the solution set quality. Next, an embedded simulation engine is used to perform virtual operating condition loading tests, obtaining simulated energy efficiency decay curves. By analyzing the overlap with preset curves, the energy efficiency performance of the motor in actual operation is verified, ensuring the reliability of the selection results. Finally, a model downgrading mechanism is triggered to eliminate unsuitable models, generating the final recommended motor model set, providing users with high-quality and highly adaptable selection results.

[0133] Similarly, in the scenario of motor selection for logistics warehousing AGV vehicles, based on the preferred model set (such as models M1 and M3) generated in the previous embodiment, the rated operating condition data of M1 (rated voltage 48V, rated current 15A, rated speed 1200r / min) is first extracted from the manufacturer's technical documents, and its efficiency distribution characteristics at 25% (efficiency 89%), 50% (92%), 75% (90%), and 100% (88%) load points are obtained. According to the user-input "high load efficiency priority" parameter, efficiency weights are dynamically allocated (25%:0.1, 50%:0.2, 75%:0.3, 100%:0.4), and the comprehensive energy efficiency score of M1 is calculated as 0.1×89+0.2×92+0.3×90+0.4×88=89.3 (preset threshold 85, those meeting the standard are retained). Subsequently, using M1's comprehensive energy efficiency score (89.3), power density (2.1kW / kg), and temperature rise characteristics (65°C@2h) as optimization targets, a Pareto front solution set (e.g., solution set includes M1, M3, and M5) is generated. If the number of solution sets exceeds the preset limit (e.g., 5), M1 (comprehensive fuzzy evaluation value 0.82) and M3 (0.78) are selected. An embedded simulation engine is used to perform a virtual operating condition loading test on M1 (simulating continuous AGV operation for 1000 hours), generating an energy efficiency decay curve (e.g., efficiency decays to 87% after 100 hours, 85% after 500 hours, and 82% after 1000 hours). The overlap with the preset curve (88% after 100 hours, 86% after 500 hours, and 83% after 1000 hours) is 92% (preset threshold 90%), meeting the standard and being retained. The final recommended set includes M1 and M3, adapting to the long-term high-efficiency operation requirements in high-temperature warehousing scenarios.

[0134] Preferably, a multi-objective particle swarm optimization algorithm is used, with comprehensive energy efficiency score, power density, and temperature rise characteristics as optimization objectives, to establish the Pareto front solution set for each preferred motor model, specifically as follows:

[0135] Initialize the particle swarm by using the comprehensive energy efficiency score, power density, and temperature rise characteristics of each preferred motor model as the three-dimensional target vector of the particles, and randomly assign initial positions and velocities.

[0136] If data loss or anomalies are detected in the three-dimensional target vector, a data repair mechanism is activated to interpolate and complete the data based on data from adjacent models.

[0137] Calculate the fitness value for each particle to construct an initial Pareto solution set; if the number of solutions is zero, adjust the target weight coefficients and recalculate the fitness values; the formula for calculating the particle fitness value is:

[0138] ;

[0139] In the formula, This represents the particle fitness value. Energy efficiency weighting coefficient; This is the power density weighting coefficient; This is the weighting coefficient for temperature rise characteristics; This is the normalized mean of the rated power; This is the normalized mean of efficiency; This is the normalized inverse mean of the temperature rise characteristics.

[0140] The particle position update algorithm is executed to dynamically adjust the particle trajectory based on the individual optimal solution and the global optimal solution. If a particle is detected to be trapped in a local optimum, a mutation operation is introduced to randomly perturb the particle position in order to escape the local extremum.

[0141] It should be noted that the difference between each particle's current position and its individual optimal solution and the global optimal solution is calculated to determine the particle's direction of motion in the target space. If the difference between the particle's position and the optimal solution is too small, the particle is considered to be trapped in a local optimum. Next, the particle's velocity vector is adjusted according to preset inertia weights, individual learning factors, and social learning factors to update the particle's position. If the particle still fails to escape the local extremum after the position update, a mutation operation is introduced to randomly perturb the particle's position parameters, causing it to generate a new exploration direction in the target space. Subsequently, the updated particle position and velocity are recorded as the basis data for the next iteration. Finally, the updated particle swarm is output as input data for subsequent non-dominated sorting and Pareto solution set construction.

[0142] The updated particles are classified using a non-dominated sorting method to select a new Pareto solution set; if the number of solutions exceeds a preset limit, crowding calculation is initiated, and solutions with uniform distribution are retained first.

[0143] It should be noted that the updated particle swarm undergoes a non-dominated sorting process, dividing the particles into multiple levels based on their dominance relationships. The first level consists of particles not dominated by any other particles, forming the initial Pareto solution set. If the number of solutions is zero, the target weight coefficients are adjusted and the fitness values ​​are recalculated. If the number of solutions exceeds a preset upper limit, the crowding distance of each particle in the target space is calculated. A larger crowding distance indicates a sparser distribution of particles in the solution set. Then, particles with larger crowding distances are preferentially retained to ensure a uniform distribution of solutions in the target space. Finally, a new Pareto solution set is output as the basis for subsequent convergence assessments. The formula for calculating the crowding distance of each particle in the target space is:

[0144] ;

[0145] In the formula, For particles Crowded distance; To optimize the target quantity; In order to achieve the goal In terms of dimensions, particles The fitness value of the adjacent particle on the right; In order to achieve the goal In terms of dimensions, particles The fitness value of the particle adjacent to the left; For the goal It is the maximum value among all particles; For the goal It is the minimum value among all particles.

[0146] Meanwhile, a solution set convergence judgment mechanism is established. If the number of consecutive iterations exceeds the threshold and the solution set change rate is lower than the preset value, the algorithm is judged to have converged; if it has not converged, iterative optimization continues.

[0147] Finally, the final Pareto front solution set is output, and the corresponding motor model and key parameters of each solution are labeled.

[0148] For example, as shown in Figure 5, Figure 5 is a trajectory diagram of a single particle. The black curve in the figure is the complete trajectory of a single particle in the three-dimensional target space; the red dot is the starting position of the particle; the blue dot is the ending position of the particle; the orange dot is the position of the global optimal solution; the three coordinate axes represent: comprehensive energy efficiency score, power density and temperature rise characteristics, respectively.

[0149] As shown in Figure 6, the particle swarm distribution and Pareto front plot are shown. The colored scatter dots represent the distribution of each particle in the three-dimensional target space; the gray surface represents the Pareto front, that is, the distribution area of ​​non-dominated solutions; the color intensity represents the fitness value of the particles; the three coordinate axes also represent three optimization objectives: comprehensive energy efficiency score, power density and temperature rise characteristics.

[0150] In summary, Figures 5 and 6 visually illustrate how a single particle gradually optimizes its position under the influence of its individual optimal solution and the global optimal solution; how the entire particle swarm is distributed in the target space and their relationship with the Pareto front; and how the algorithm finds the optimal balance point among the three objectives. These figures fully demonstrate the working principle and optimization process of the multi-objective particle swarm optimization algorithm in motor selection.

[0151] It should be noted that an initial Pareto solution set is constructed using fitness value calculation and non-dominated sorting. A particle position update algorithm is then used to dynamically optimize particle trajectories, ensuring both optimality and diversity of the solution set. If a particle gets stuck in a local optimum, a mutation operation is introduced to escape local extrema, improving the global search capability of the solution. Next, a uniformly distributed solution set is selected through crowding calculation to ensure broad coverage of the target space. Simultaneously, a convergence judgment mechanism is established to ensure the algorithm terminates promptly upon reaching the optimal solution, avoiding ineffective iterations. Finally, the final Pareto front solution set is output, along with annotations of the corresponding motor model and key parameters for each solution, providing users with high-quality, multi-dimensional selection references.

[0152] Similarly, in the scenario of motor selection for AGVs in logistics warehousing, based on the preferred model set selected in the previous embodiment (e.g., model M1: comprehensive energy efficiency score 89.3, power density 2.1kW / kg, temperature rise 65°C), when initializing the particle swarm, the three-dimensional target vector of M1 and its similar preferred models ([89.3, 2.1, 65]) is used as the initial position of the particles, and the initial velocity is randomly assigned (e.g., [0.2, -0.1, 0.3]). If the temperature rise characteristic data of a certain model is detected to be missing (e.g., model M3 is missing temperature rise data), it is filled in by interpolation based on the data of the adjacent models M1 and M5 (M1 temperature rise 65°C, M5 temperature rise 68°C, average 66.5°C). When calculating the fitness value, the user-defined weights (μ=0.5, ρ=0.3, τ=0.2) are substituted into the formula F=0.5×0.89 (normalized energy efficiency value)+0.3×0.72 (normalized power density value)+0.2×(1-0.65)=0.763. When performing particle position updates, if the difference between the individual optimal solution (F=0.763) and the global optimal solution (F=0.78) of M1 is less than a threshold (e.g., 5%), a mutation operation is triggered, randomly perturbing its power density parameter (e.g., adjusting it from 0.72 to 0.75), and generating a new position [89.3, 2.25, 65]. An initial Pareto solution set (containing M1, M3, and M5) is selected through non-dominated sorting. If the number of solutions exceeds a preset limit (e.g., 5), the crowding distance is calculated (e.g., the adjacent particle spacing of M1 in the energy efficiency dimension is 0.15, in the power density dimension it is 0.12, in the temperature rise dimension it is 0.08, and the total crowding distance CD = 0.35), prioritizing the retention of models with high CD values. Convergence is determined when the solution set change rate is less than 2% for 10 consecutive iterations. The final Pareto front solution set is output (as shown in Figure 6, M1 and M5 are distributed in the gray curved surface area), with key parameters labeled (e.g., M1: energy efficiency 89.3, power density 2.1kW / kg, temperature rise 65°C) to adapt to the high-efficiency, long-term operation requirements of AGVs.

[0153] In summary, to address the challenges of simultaneously optimizing multiple objective parameters such as energy efficiency, power density, and temperature rise characteristics during motor selection, and the tendency of traditional methods to fall into local optima and uneven solution distribution, this invention employs a multi-objective particle swarm optimization algorithm to dynamically balance various optimization objectives, automatically generating a uniformly distributed Pareto optimal solution set. Data repair and mutation operations ensure the integrity of the solution and global search capability; non-dominated sorting and congestion calculation maintain the diversity of the solution set. The final output is a set of motor models that achieve an optimal balance of comprehensive energy efficiency, power density, and temperature rise characteristics, providing a scientific basis for multi-objective decision-making.

[0154] Preferably, if the number of solutions to the Pareto front exceeds a preset upper limit, a fuzzy comprehensive evaluation method is introduced to perform a secondary screening of the solutions until the number of solutions to the Pareto front falls below the preset upper limit, specifically as follows:

[0155] Extract the comprehensive energy efficiency score, power density, and temperature rise characteristics of each motor model from the Pareto front solution set, and construct a multi-dimensional evaluation matrix;

[0156] Based on the energy efficiency priority parameters input by the user, the weight coefficients of each evaluation indicator are dynamically allocated; if the weight allocation result deviates from the historical preferred mode by more than the preset deviation value, the weight adjustment mechanism is triggered to reallocate the weight values.

[0157] It should be noted that the process involves parsing the user-inputted energy efficiency priority parameters and extracting key indicators such as high-load efficiency priority, average efficiency priority, or low-load efficiency priority. Based on a preset energy efficiency priority-weight mapping table, initial weight values ​​for each evaluation indicator (such as comprehensive energy efficiency score, power density, and temperature rise characteristics) are dynamically assigned according to the extracted key indicators. Then, the initial weight values ​​are normalized to ensure that the sum of the weights of each evaluation indicator is 1. Subsequently, the final weight values ​​are written into the dynamic weight rule base, and a weight allocation report is generated for user confirmation. If the user requests modifications, the weight allocation process is re-executed. Finally, the complete weight allocation process and result log are recorded as a reference for subsequent weight rule optimization.

[0158] The preset energy efficiency priority-weight mapping table is a predefined rule lookup table used to automatically match and allocate the weight coefficients of efficiency evaluation indicators for each load point according to the energy efficiency priority specified by the user (such as high load efficiency priority, average efficiency priority, etc.), ensuring that the weight allocation accurately corresponds to the user's needs.

[0159] Establish a fuzzy membership function to normalize each evaluation index and eliminate dimensional differences; if uneven distribution of the normalized data is detected, adjust the membership function parameters.

[0160] It should be noted that, firstly, based on the data distribution characteristics of each evaluation indicator (such as energy efficiency score and power density), a triangular or trapezoidal membership function type is selected, and initial parameters (such as triangle vertex positions and trapezoidal boundary values) are set; the data is then normalized (e.g., mapping 0-100% efficiency to the [0,1] interval). If uneven data distribution is detected after normalization (e.g., energy efficiency scores are concentrated in the high-end range of 0.8-1.0, while temperature rise data are scattered in 0.3-0.9), the membership function parameters are adjusted: for dense intervals (e.g., energy efficiency 0.8-1.0), the triangle vertex spacing is reduced to improve resolution; for scattered intervals (e.g., temperature rise 0.3-0.9), the trapezoidal boundary range is expanded to cover a wider data domain, and the distribution curve is refitted. For example, the temperature rise membership function is expanded from the original trapezoidal boundary [0.5,0.7] to [0.3,0.9] to ensure uniform membership of data points. After adjustment, the function fitting effect is verified until the normalized data distribution approaches uniformity.

[0161] The weighted average method is used to calculate the comprehensive fuzzy evaluation value of each motor model, and the solution set is sorted according to the comprehensive fuzzy evaluation value;

[0162] It should be noted that the normalized weight coefficients of each evaluation indicator (such as energy efficiency weight 0.5, power density 0.3, and temperature rise 0.2) are obtained from the dynamic weight rule base. The normalized evaluation indicator data for each motor model (such as energy efficiency 0.9, power density 0.7, and temperature rise 0.8 for a certain model) are extracted. Each indicator value is multiplied by its corresponding weight (0.9×0.5+0.7×0.3+0.8×0.2=0.82), and the results are accumulated to obtain the comprehensive fuzzy evaluation value. The solution set is sorted from high to low according to the evaluation value. If the evaluation values ​​are the same, they are sorted a second time according to the energy efficiency score. Finally, a sorted list of preferred models is generated for users to make decisions.

[0163] Set an upper limit threshold for the number of solution sets, and sequentially remove the motor models with the lowest comprehensive fuzzy evaluation values ​​until the number of motor models after removal is less than the upper limit threshold for the number of solution sets; if the number of solution sets after removal still exceeds the upper limit, gradually increase the removal criteria until the quantity requirement is met.

[0164] Finally, the filtered Pareto front solution set is output, and the comprehensive fuzzy evaluation value and key parameters of each model are labeled.

[0165] Similarly, in the scenario of motor selection for AGVs in logistics warehousing, for the Pareto front solution set generated in the previous embodiment (e.g., containing 10 candidate models such as M1, M3, and M5, with a preset upper limit of 5), the key parameters of each model are first extracted (e.g., M1: comprehensive energy efficiency score 89.3, power density 2.1kW / kg, temperature rise 65°C) to construct a multi-dimensional evaluation matrix. Based on the user-input parameter of "high load efficiency priority", weight coefficients are dynamically allocated (energy efficiency score 0.5, power density 0.3, temperature rise 0.2); if it is detected that the temperature rise weight is usually 0.15 in the historical preferred mode (the current deviation of 0.2 exceeds the preset threshold by 15%), the weight adjustment mechanism is triggered to correct the temperature rise weight to 0.15, and adjust the energy efficiency and power density to 0.55 and 0.3, respectively. Subsequently, a triangular membership function is established to normalize the parameters (e.g., energy efficiency score is mapped to [0.8, 1.0], power density to [0.6, 0.9], and temperature rise is inversely mapped to [0.7, 1.0]). If the detected temperature rise data is concentrated in the 60-70°C range (normalized to 0.6-0.8), the membership function parameters are adaptively adjusted to expand the coverage range to 0.5-0.9. A weighted average method is used to calculate the comprehensive fuzzy evaluation value (e.g., M1 score: 0.55×0.893+0.3×0.7+0.15×0.8=0.801). After sorting by score, the bottom 5 are eliminated (e.g., models with scores < 0.7). If the remaining 7 still exceed the upper limit (e.g., the upper limit is 5), the elimination standard is increased to a score < 0.75. Finally, the 5 filtered models are output (e.g., M1: 0.801, M5: 0.78), and key parameters are labeled to adapt to the high energy efficiency requirements of AGVs.

[0166] In summary, this invention dynamically adjusts evaluation weights based on user needs to ensure that the screening results conform to actual application scenarios; it eliminates dimensional differences through an adaptive membership function to improve evaluation accuracy; and it adopts a quantitative scoring mechanism to achieve precise sorting and screening of the solution set, ultimately outputting a set of motor models with controllable quantity and optimal quality, thereby improving selection efficiency and decision reliability.

[0167] Preferably, the recommended motor models are divided into multiple motor selection levels, specifically:

[0168] Extract the comprehensive energy efficiency score, cost-effectiveness index and compatibility parameters of all models in the recommended motor model set, and construct a three-dimensional classification feature matrix;

[0169] Based on the core performance thresholds set by the user, a set of models with comprehensive energy efficiency scores higher than the benchmark and compatibility standards are selected and marked as the primary recommended candidate set; if the candidate set is empty, the threshold range is automatically widened and the adjustment parameters are recorded.

[0170] For the remaining models, perform an economic analysis, calculate the life cycle cost-benefit ratio, and screen out models with a cost-benefit ratio lower than the industry average and compatibility parameters within the acceptable range, classifying them as economic substitutes; if missing cost data is detected, call the historical procurement database to fill in the substitute values;

[0171] For models not selected as the primary recommended or economical alternatives, extended compatibility verification will be conducted, and models that pass the extended compatibility verification will be designated as extended compatible models.

[0172] It should be noted that "extended compatibility verification" refers to conducting cross-platform and cross-scenario applicability assessments on motor models that were not selected as primary recommended or economical alternatives. This involves simulating tests to verify their adaptability to different operating environments, load types, and control systems, thus validating their compatibility and flexibility in diverse application scenarios. Specifically, this includes steps such as interface compatibility analysis, load fluctuation tolerance testing, and multi-condition operational stability verification. This ensures that these models can meet the usage requirements of non-core application scenarios or special needs scenarios, thereby providing users with more comprehensive and flexible selection options.

[0173] Similarly, in the scenario of motor selection for AGVs in logistics warehousing, based on the recommended motor model set generated in the previous embodiment (e.g., including models M1, M3, and M5), the key parameters of each model are first extracted to construct a three-dimensional classification feature matrix (e.g., M1: comprehensive energy efficiency score 89.3, cost-effectiveness index 0.85, compatibility score 92). According to the core performance thresholds set by the user (energy efficiency ≥ 85 points, compatibility ≥ 90), M1 (89.3 / 92) is selected and marked as the primary recommended model; when the candidate set is empty, the energy efficiency threshold is automatically relaxed to 80 points. Economic analysis is performed on the remaining models M3 (energy efficiency 82 / cost-effectiveness 0.78) and M5 (energy efficiency 79 / cost-effectiveness 0.72), and the life cycle cost-effectiveness ratio is calculated (industry average 0.8), classifying M3 as an economic substitute. Extended compatibility verification is performed on M5, and its stability (efficiency fluctuation < 5%) in a cold chain environment (-20°C) and a frequency converter control system is tested through simulation, confirming it as an extended compatible model. The final output is a hierarchical selection result: the primary recommended type M1 (highly efficient for ambient temperature storage), the economical alternative M3 (balanced cost-effectiveness), and the extended compatibility type M5 (for special scenarios), to meet the needs of different AGV deployment scenarios.

[0174] In summary, this step, through a systematic and multi-level process, enhances the scientific rigor, accuracy, and practicality of motor selection, providing users with high-quality, multi-dimensional support for selection decisions.

[0175] The selection criteria parameters include gross weight under full load, speed under full load, acceleration under full load, diameter of traveling wheels / drum, energy efficiency level requirements, three-dimensional size limitations of installation space, energy efficiency priority parameters, environmental temperature and humidity gradient parameters, protection level code, expected life index, and application scenario characteristics.

[0176] As shown in Figure 3, the second aspect of the present invention discloses a motor selection system 6, which includes a memory 41 and a processor 52. The memory 41 stores a motor selection method program. When the motor selection method program is executed by the processor 52, the steps of any of the motor selection methods described in the present invention are implemented.

[0177] The above are merely specific embodiments of the present invention, but the scope of protection of the present invention is not limited thereto. Any changes or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in the present invention should be included within the scope of protection of the present invention.

Claims

1. A method for selecting a motor, characterized in that, The process includes the following steps: obtaining the selection benchmark parameters input by the user, which include the total weight under full load, speed under full load, acceleration, diameter of the traveling wheel / drum, energy efficiency level requirements, three-dimensional size limitations of the installation space, energy efficiency priority parameters, environmental temperature and humidity gradient parameters, protection level code, expected life index, and application scenario characteristics; The selection benchmark parameters are converted into a benchmark measurement system to obtain standardized feature vectors. Specifically, this includes: extracting the numerical values ​​and unit identifiers of each input selection benchmark parameter, and dynamically constructing a parameter-unit mapping table; identifying the unit identifier of each parameter in the mapping table, and if the unit is imperial or non-standard metric, converting the corresponding selection benchmark parameter into an equivalent value under the benchmark unit; for the converted selection benchmark parameters, performing interval mapping on each parameter according to a preset motor performance parameter threshold range, and concatenating the normalized numerical parameters and the encoded discrete parameters according to weight priority through a feature fusion layer to form a multi-dimensional standardized feature vector; and constructing a distributed data... According to the query database, the standardized feature vector is imported into a distributed data query database to match a set of candidate motor models; the comprehensive performance score of each candidate motor model is obtained, and a model elimination mechanism is triggered based on a preset score threshold. The remaining candidate motor models are sorted by performance parameters to generate a set of preferred motor models. Specifically, obtaining the comprehensive performance score of each candidate motor model includes: extracting key performance parameters of each model from the candidate motor model set, including rated power curves, efficiency distribution maps, and temperature rise characteristic data, and constructing a three-dimensional performance matrix; and automatically assigning torque response coefficients, energy efficiency priority factors, and environmental adaptability weights according to the application scenario characteristics input by the user. A dynamic weight allocation rule is formed; based on the dynamic weight rule, the parameters in the three-dimensional performance matrix are weighted and calculated to generate a comprehensive performance score for each candidate motor model; the comprehensive energy efficiency score of each preferred motor model is obtained, and a Pareto front solution set is established for secondary screening to obtain a recommended set of motor models. Specifically, establishing the Pareto front solution set for secondary screening includes: using a multi-objective particle swarm optimization algorithm, with comprehensive energy efficiency score, power density, and temperature rise characteristics as optimization objectives, to establish the Pareto front solution set for each preferred motor model; and calling an embedded simulation engine to perform virtual operating condition loading tests on each preferred motor model based on the Pareto front solution set. The process involves: obtaining simulated energy efficiency degradation curves for each preferred motor model; comparing these curves with preset energy efficiency degradation curves to determine the overlap between them; if the overlap between a preferred motor model's simulated energy efficiency degradation curve and the preset energy efficiency degradation curve is not greater than the preset overlap, a model downgrade mechanism is triggered, removing the model from the preferred queue; obtaining the remaining preferred motor models in the preferred motor model set to obtain a recommended motor model set; dividing the recommended motor models into multiple motor selection levels and outputting these levels as the final motor selection result.

2. The motor selection method according to claim 1, characterized in that, Construct a distributed data query library, import the standardized feature vector into the distributed data query library to match the candidate motor model set, specifically: obtain the baseline parameter data stream of motors from each manufacturer, convert the baseline parameter data stream of motors from each manufacturer into standardized parameters with unified dimensions through a preset motor industry standard terminology mapping table, and construct a standardized parameter table; The standardized parameter table is matrixed to form a multidimensional feature matrix, where each row represents a motor model and each column corresponds to a standardized parameter feature. A data query library is constructed by importing the multidimensional feature matrix into the data query library to obtain a distributed data query library. The standardized feature vectors are dimensionally expanded to match the column dimensions of the multidimensional feature matrix to form a query vector matrix. The query vectors are compared row by row with each row in the feature matrix, and the Euclidean distance between them on each feature dimension is calculated. If the Euclidean distance of a row is less than a preset distance threshold, the motor model in the corresponding row is marked as a candidate model. The marked candidate model set is prioritized and sorted according to the Euclidean distance from smallest to largest to generate the final candidate motor model set. Throughout the process, if the set of candidate motor models is empty, the preset distance threshold will be automatically adjusted or the user will be prompted to re-enter the selection benchmark parameters.

3. The motor selection method according to claim 1, characterized in that, The process of ranking the retained candidate motor models by performance parameters to generate a preferred set of motor models involves: comparing the comprehensive performance score of each candidate motor model with a preset score threshold; if the comprehensive performance score of a candidate motor model is less than the preset score threshold, a model elimination mechanism is triggered to remove the candidate motor model from the candidate motor model set; the retained candidate models are then sorted from largest to smallest by their comprehensive performance scores to generate a preferred set of motor models; if there are cases where the scores are the same, a secondary sort is performed based on the energy efficiency priority factor to ensure the uniqueness of the sorting results, and the preferred sequence is used as input data for the next stage of load characteristic matching verification.

4. The motor selection method according to claim 1, characterized in that, A multi-objective particle swarm optimization algorithm is adopted, with comprehensive energy efficiency score, power density, and temperature rise characteristics as optimization objectives. Pareto front solution sets for each preferred motor model are established. Specifically: the particle swarm is initialized by using the comprehensive energy efficiency score, power density, and temperature rise characteristics of each preferred motor model as the three-dimensional objective vector of the particles, and initial positions and velocities are randomly assigned; the fitness value of each particle is calculated, and an initial Pareto solution set is constructed; if the number of solutions is zero, the objective weight coefficients are adjusted and the fitness values ​​are recalculated; a particle position update algorithm is executed to dynamically adjust the particle trajectories based on the individual optimal solution and the global optimal solution. The algorithm iterates through a series of steps. If a particle is detected to be trapped in a local optimum, a mutation operation is introduced to randomly perturb the particle's position to escape the local extremum. The updated particles are then classified using a non-dominated sorting method to select a new Pareto solution set. If the number of solutions exceeds a preset upper limit, crowding calculation is initiated, prioritizing the retention of uniformly distributed solutions. Simultaneously, a convergence judgment mechanism for the solution set is established. If the number of consecutive iterations exceeds a threshold and the rate of change of the solution set is lower than a preset value, the algorithm is considered to have converged. If it has not converged, iterative optimization continues. Finally, the final Pareto front solution set is output, and the corresponding motor model and key parameters of each solution are labeled.

5. The motor selection method according to claim 4, characterized in that, If the number of Pareto front solutions exceeds a preset upper limit, a fuzzy comprehensive evaluation method is introduced to perform a secondary screening of the solutions until the number of Pareto front solutions falls below the preset upper limit. Specifically, this involves: extracting the comprehensive energy efficiency score, power density, and temperature rise characteristic data of each motor model from the Pareto front solutions to construct a multi-dimensional evaluation matrix; dynamically allocating the weight coefficients of each evaluation index according to the energy efficiency priority parameters input by the user; establishing a fuzzy membership function to normalize each evaluation index and eliminate dimensional differences; if uneven distribution of the normalized data is detected, adjusting the membership function parameters; calculating the comprehensive fuzzy evaluation value of each motor model using a weighted average method and sorting the solutions based on the comprehensive fuzzy evaluation value; setting an upper limit threshold for the number of solutions, and sequentially removing the motor models with the lowest comprehensive fuzzy evaluation values ​​until the number of motor models after removal is less than the upper limit threshold for the number of solutions. If the number of solutions after elimination still exceeds the upper limit, the elimination criteria are gradually increased until the quantity requirement is met. Finally, the Pareto front solution set after filtering is output, and the comprehensive fuzzy evaluation value and key parameters of each model are labeled.

6. The motor selection method according to claim 1, characterized in that, The process involves obtaining comprehensive energy efficiency scores for each preferred motor model and establishing Pareto front solution sets for secondary screening to obtain a recommended set of motor models. This process also includes: obtaining rated operating condition data for each preferred motor model; extracting efficiency distribution characteristics of each preferred motor model under different load points based on the rated operating condition data; dynamically allocating efficiency weight coefficients according to user-inputted energy efficiency priority parameters; and calculating the comprehensive energy efficiency score of each preferred motor model based on its efficiency distribution characteristics under different load points. If the comprehensive energy efficiency score of a preferred motor model is lower than a preset score threshold, a model downgrade mechanism is triggered, removing the model from the preferred queue. If the number of Pareto front solution sets exceeds a preset upper limit, a fuzzy comprehensive evaluation method is introduced to perform secondary screening of the solution sets until the number of Pareto front solution sets falls below the preset upper limit.

7. The motor selection method according to claim 1, characterized in that, The recommended motor models are divided into multiple motor selection levels, specifically: The comprehensive energy efficiency score, cost-effectiveness index, and compatibility parameters of all models in the recommended motor model set are extracted to construct a three-dimensional classification feature matrix; based on the user-defined core performance threshold, a set of models with comprehensive energy efficiency scores higher than the benchmark value and meeting compatibility standards is selected and marked as the primary recommended candidate set; if the candidate set is empty, the threshold range is automatically widened and the adjustment parameters are recorded; economic analysis is performed on the remaining models to calculate the life-cycle cost-effectiveness ratio, and models with cost-effectiveness ratios lower than the industry average and compatibility parameters within the allowable range are selected and classified as economic substitutes; if missing cost data is detected, the historical procurement database is used to fill in the substitute values; extended compatibility verification is performed on models not selected as the primary recommended or economic substitute types, and models that pass the extended compatibility verification are designated as extended compatible types.

8. A motor selection system, characterized in that, The motor selection system includes a memory and a processor. The memory stores a motor selection method program. When the motor selection method program is executed by the processor, it implements the steps of the motor selection method as described in any one of claims 1 to 7.

Images