A large-flow centrifugal fan impeller optimization design method

Abstract

The present application belongs to the field of centrifugal fan optimization design, and relates to a large-flow centrifugal fan impeller optimization design method. The method obtains impeller optimization design input data and performs consistency preprocessing to obtain a design data set; extracts a candidate working condition set based on the design data set; constructs an optimizable impeller parameter set and an outlet flow field constraint set based on the candidate working condition set; performs impeller parameter evaluation on the optimizable impeller parameter set combined with the outlet flow field constraint set to obtain an impeller comprehensive evaluation result; and performs optimization solving processing on the impeller comprehensive evaluation result to obtain a target impeller parameter result. The method is conducive to taking into account large air volume efficiency, wide working condition stability and end air supply uniformity.

Description

Technical Field

[0001] This invention belongs to the field of centrifugal fan impeller design technology, specifically relating to an optimized design method for a high-flow centrifugal fan impeller. Background Technology

[0002] In recent years, with the continuous high-density deployment of computing equipment in artificial intelligence data centers, the heat dissipation load of server racks has exhibited rapid fluctuations and localized concentrations. High-flow-rate centrifugal fans used in precision air conditioning units, indirect evaporative cooling units, and related air supply equipment not only need to meet the air supply requirements under rated airflow conditions but also need to maintain impeller aerodynamic stability, outlet flow field distribution stability, and terminal air supply adaptability under different heat load levels. In these application scenarios, impeller geometry parameters directly affect the fan's static pressure efficiency, pressure pulsation, noise level, and unit outlet airflow distribution. Therefore, how to optimize impeller design for wide load variations has become a crucial issue in the design of related equipment.

[0003] In existing technologies, the basic structure of the centrifugal impeller is typically determined based on the unit's rated air volume, rated static pressure, and installation boundaries. Then, using methods such as numerical simulation, multi-objective optimization, or parameter optimization, geometric parameters such as blade installation angle, blade wrap angle, thickness distribution, and number of blades are adjusted to improve efficiency, noise, or strength performance under design conditions. After the equipment is put into operation, speed regulation is used to adapt to different loads. This type of technical approach has become a relatively mature application method for performance optimization near a single design point.

[0004] However, the above solutions still have two shortcomings when facing application environments with significant heat load fluctuations, such as artificial intelligence data centers. First, existing impeller designs are usually based on parameter optimization at a single design point or a few steady-state conditions, failing to effectively transform the load fluctuation characteristics during operation into multi-condition constraints in the impeller design stage. This leads to phenomena such as enhanced blade separation, increased backflow, outlet velocity distribution shift, and increased noise under medium-low load or rapidly changing load conditions, making it difficult to balance rated performance and wide-condition operational stability. Second, existing design objectives focus more on the efficiency, noise, or structural indicators of the fan body, failing to further transform the influence of terminal hotspot sensitive areas, regional pressure difference changes, and local backflow into constraints on the impeller outlet flow field. Therefore, even when the overall air volume meets the requirements, problems such as insufficient air supply in local hotspot areas or uneven air supply distribution may still occur, making it difficult to simultaneously balance fan body performance and terminal air supply uniformity. To address this, this invention proposes an optimized design method for high-flow centrifugal fan impellers. Summary of the Invention

[0005] In view of the aforementioned existing problems, the present invention is proposed.

[0006] This invention provides a method for optimizing the design of impellers for high-flow centrifugal fans. The aim is to at least partially solve one or more of the problems in existing impeller optimization, which often revolve around a single design point and make it difficult to balance the stability of operation under wide operating conditions and the uniformity of air delivery at the end. At the same time, it is difficult to move the sensitive area of ​​hot spots at the end, pressure fluctuations, and backflow effects forward as constraints on the impeller outlet flow field, resulting in one or more deviations between the optimization results and the actual air delivery requirements.

[0007] To address the above problems, the present invention provides the following technical solution:

[0008] A method for optimizing the design of a high-flow-rate centrifugal fan impeller includes:

[0009] Obtain the impeller optimization design input data and perform consistency preprocessing to obtain the design data set;

[0010] Candidate operating conditions are extracted from the design data set to obtain a candidate operating condition set;

[0011] The impeller parameters are constructed from the candidate operating condition set to obtain an optimizable impeller parameter set;

[0012] The outlet flow field constraint is constructed by performing an outlet flow field constraint on the candidate working condition set to obtain the outlet flow field constraint set;

[0013] The impeller parameter evaluation is performed on the optimizable impeller parameter set and the outlet flow field constraint set to obtain the comprehensive impeller evaluation result;

[0014] The comprehensive evaluation results of the impeller are optimized and solved to obtain the target impeller parameters.

[0015] In a preferred embodiment, the step of extracting candidate operating conditions from the design data set to obtain a candidate operating condition set includes:

[0016] The design data set is sliced ​​according to a unified time window to obtain a set of working condition segments;

[0017] Interval statistics are performed on the air volume data, static pressure data, and temperature rise data in the set of operating condition segments to obtain operating condition characteristic records;

[0018] Representative working conditions are selected from the working condition feature records to obtain a candidate working condition set.

[0019] As a preferred embodiment, the representative working condition screening of the working condition feature records includes:

[0020] The load fluctuation spectrum is obtained by performing fluctuation statistics on the load change amplitude, duration, and frequency of occurrence in the aforementioned operating condition characteristic records.

[0021] Clustering and merging the fluctuation intervals in the load fluctuation spectrum yields representative operating condition categories.

[0022] The weights of representative working condition categories driven by load fluctuation spectrum are calculated by a multi-condition design weight generation algorithm to determine the multi-condition design weights of each candidate working condition in the candidate working condition set.

[0023] As a preferred embodiment, the step of constructing impeller parameters from the candidate operating condition set to obtain an optimizable impeller parameter set includes:

[0024] Read the reference impeller parameters and installation boundary information from the design data set;

[0025] Perform impeller design boundary transformation on each candidate operating condition in the candidate operating condition set to obtain impeller design boundary parameters;

[0026] The impeller geometry parameters are expanded by the reference impeller parameters, the installation boundary information, and the impeller design boundary parameters to obtain a set of optimizable impeller parameters.

[0027] As a preferred embodiment, the step of constructing an outlet flow field constraint set from the candidate operating condition set includes:

[0028] Read the terminal temperature rise deviation data, regional pressure difference fluctuation data, and reflux characterization data corresponding to the candidate operating condition set from the design data set;

[0029] Sensitivity correlation analysis was performed on the terminal temperature rise deviation data, the regional pressure difference fluctuation data, and the reflux characterization data to obtain the hot spot sensitivity matrix;

[0030] The hot spot sensitivity matrix is ​​subjected to outlet partition mapping and constraint quantization to obtain an outlet flow field constraint set containing non-uniform penalty constraints for each outlet partition.

[0031] As a preferred embodiment, the step of performing outlet partitioning mapping and constraint quantization on the hotspot sensitivity matrix includes:

[0032] The sensitivity values ​​of each region in the hot spot sensitivity matrix are matched with the corresponding partitions of the fan outlet annular surface to obtain the partition sensitivity records.

[0033] The non-uniform penalty constraint generation algorithm for hot spot sensitivity inverse mapping is used to calculate the penalty value of the partition sensitivity record to obtain the non-uniform penalty constraint corresponding to each exit partition.

[0034] The non-uniform penalty constraints corresponding to each exit partition are combined and encapsulated to complete the construction of the exit flow field constraint set containing the non-uniform penalty constraints of each exit partition.

[0035] As a preferred embodiment, the step of performing impeller parameter evaluation on the optimizable impeller parameter set and the outlet flow field constraint set to obtain a comprehensive impeller evaluation result includes:

[0036] Aerodynamic simulations were performed on each impeller parameter combination in the optimizable impeller parameter set under each candidate working condition to obtain the working condition aerodynamic performance record.

[0037] The static pressure efficiency, outlet flow field uniformity, low load separation risk, and blade frequency noise proxy value are calculated from the aerodynamic performance records of the aforementioned operating conditions to obtain the sub-operating condition evaluation records.

[0038] The evaluation records for each working condition are weighted and summarized according to the design weights of the multi-working-condition system, and then comprehensively evaluated in conjunction with the non-uniform penalty constraints of each outlet zone in the outlet flow field constraint set to obtain the comprehensive evaluation result of the impeller.

[0039] As a preferred implementation, the weighted summarization of the evaluation records for each working condition according to the multi-working-condition design weights, and the comprehensive evaluation combined with the non-uniform penalty constraints of each outlet zone in the outlet flow field constraint set, includes:

[0040] The evaluation records for each working condition are weighted and summarized according to the multi-condition design weights corresponding to each candidate working condition to obtain a weighted evaluation record.

[0041] The weighted evaluation record is penalized and corrected according to the non-uniform penalty constraint of each outlet partition in the outlet flow field constraint set to obtain the corrected evaluation record.

[0042] The revised evaluation records are comprehensively sorted to determine the overall evaluation result of the impeller.

[0043] As a preferred embodiment, the optimization solution process performed on the comprehensive evaluation results of the impeller to obtain the target impeller parameter results includes:

[0044] The candidate parameter sequence is obtained by prioritizing the various impeller parameter combinations in the comprehensive evaluation results of the impeller.

[0045] The candidate parameter sequence is subjected to installation boundary compliance verification and outlet flow field constraint compliance verification to obtain a feasible parameter sequence;

[0046] The feasible parameter sequence is converged and iterated to obtain the target impeller parameter results.

[0047] As a preferred embodiment, prioritizing the combinations of impeller parameters in the comprehensive impeller evaluation result includes:

[0048] The various impeller parameter combinations in the comprehensive evaluation results of the impeller are ranked in a coordinated manner according to the static pressure efficiency index, the outlet flow field uniformity index, the low load separation risk index, and the blade frequency noise proxy index to obtain a coordinated optimization candidate sequence;

[0049] The collaborative optimization candidate sequence is subjected to wide-condition stability verification and installation boundary compliance verification to obtain a feasible parameter sequence;

[0050] The feasible parameter sequence is iteratively solved using an aerothermal field noise co-optimization solution algorithm to obtain the target impeller parameter results.

[0051] Compared with the prior art, the present invention has the following beneficial effects:

[0052] 1. By extracting the load change process during unit operation into a set of candidate operating conditions, and further generating multi-condition design weights corresponding to each candidate operating condition, the evaluation of impeller parameters is no longer limited to a single design operating condition, but can simultaneously take into account the evaluation results corresponding to high load continuous operating conditions, load switching operating conditions, and medium- and low load sensitive operating conditions; the target impeller parameter results obtained in this way have a more complete operating condition coverage basis in a wide range of operating conditions.

[0053] 2. Sensitivity correlation analysis is performed on the terminal temperature rise deviation, regional pressure difference fluctuation, and recirculation characterization data. The obtained hot spot sensitivity matrix is ​​then mapped inversely to the non-uniform penalty constraints corresponding to each outlet zone. This allows the thermal risk distribution in the terminal region to be shifted forward to the impeller outlet flow field constraint construction process. Accordingly, in the impeller parameter evaluation stage, the velocity deviation of the outlet zone can be constrained and corrected, so that the target impeller parameter results simultaneously take into account the fan body performance and the terminal air supply distribution requirements.

[0054] 3. Static pressure efficiency, outlet flow field uniformity, low-load separation risk, blade frequency noise proxy value, and outlet flow field constraint compliance are all incorporated into the comprehensive evaluation results of the impeller. In the optimization solution process, the impeller parameter combination is iteratively screened by combining wide-condition stability verification and installation boundary compliance verification. This transforms the selection of target impeller parameters from a single index to a collaborative determination result under multiple constraints. Accordingly, it is beneficial to ensure that the obtained impeller parameter results maintain good feasibility and parameter adaptability in engineering applications. Attached Figure Description

[0055] Figure 1 An exemplary flowchart of a high-flow-rate centrifugal fan impeller optimization design method provided in an embodiment of the present invention.

[0056] Figure 2 A comparison diagram showing the effects of a high-flow-rate centrifugal fan impeller optimization design method provided in this embodiment of the invention with existing technologies. Detailed Implementation

[0057] To make the technical means, creative features, and achieved objectives and effects of this invention easier to understand, the invention is further described below with reference to specific embodiments. However, the following embodiments are merely preferred embodiments of this invention and not all of them. Other embodiments obtained by those skilled in the art based on the embodiments described herein without creative effort are all within the protection scope of this invention. Unless otherwise specified, the experimental methods in the following embodiments are conventional methods, and the materials and reagents used in the following embodiments are commercially available unless otherwise specified.

[0058] Example 1 combined Figure 1 The following is an exemplary flowchart of an optimized design method for a high-flow-rate centrifugal fan impeller. The specific implementation steps are as follows:

[0059] Obtain the impeller optimization design input data and perform consistency preprocessing to obtain the design data set;

[0060] The consistency preprocessing refers to the process of performing time window alignment, duplicate data processing, outlier removal, and unit unification on the input data of impeller optimization design. It is used to convert original input data from different sources, with different time calibers, and different measurement calibers into a unified set of design data that can be called.

[0061] Specifically, first, the target unit identifier, target area identifier, and target design period are determined. Based on the target unit identifier, an operation parameter reading command is sent to the unit operation monitoring terminal to read the airflow data, static pressure data, and temperature rise data recorded chronologically within the target design period. Next, based on the target area identifier, a thermal field parameter reading command is sent to the terminal environmental monitoring terminal to read the terminal temperature rise deviation data, regional pressure difference fluctuation data, and recirculation characterization data corresponding to the target design period. Finally, a design parameter reading command is sent to the impeller design parameter storage terminal to read the reference impeller parameters and installation boundary information corresponding to the target unit. For each data item read, the corresponding unit identifier, area identifier, acquisition time, and parameter unit are appended to form the impeller optimization design input data.

[0062] After acquiring the input data for the impeller optimization design, the input data is preprocessed for consistency. Specifically, the acquisition times of various data are merged according to a unified time window. Data falling within the same time window and corresponding to the same unit identifier and the same area identifier are grouped into the same window data group. When multiple records of the same parameter exist within the same window data group, a single window parameter value is generated according to a preset statistical caliber. Among them, the air volume data, static pressure data, and temperature rise data use the average value of the valid records within the window; the terminal temperature rise deviation data uses the average deviation value of the valid records within the window; the regional pressure difference fluctuation data uses the difference between the upper and lower pressure difference values ​​within the window; and the recirculation characterization data uses the recirculation event data within the window. The proportion of occurrence of items; when there are records with duplicate parameter numbers in the same window data group, retain the record with the later acquisition time and delete the other duplicate records; when any of the following is missing in a window data group: air volume data, static pressure data, temperature rise data, terminal temperature rise deviation data, regional pressure difference fluctuation data, reflux characterization data, reference impeller parameters, or installation boundary information, write a missing mark to the window data group; when a parameter value exceeds the corresponding equipment range or design boundary registration range, perform anomaly rejection on the parameter value; when the same type of parameter is recorded using different units of measurement, complete the unit unification according to a unified conversion standard.

[0063] After completing time window merging, window parameter generation, duplicate record deletion, missing record marking, anomaly removal, and unit unification, an integrity check is performed on each window data group. Data groups that simultaneously contain airflow parameters, static pressure parameters, temperature rise parameters, terminal temperature rise deviation parameters, regional pressure difference fluctuation parameters, recirculation characterization parameters, reference impeller parameters, and installation boundary information, and do not have missing record markings, are added to the design data set. Data groups with missing record markings, incomplete parameter items after anomaly removal, or failed unit unification are excluded from the design data set. Through the above processing, a design data set organized according to a unified time window and unified parameter caliber is obtained; the impeller optimization design input data...

[0064] Candidate operating conditions are extracted from the design data set to obtain a candidate operating condition set;

[0065] The candidate operating condition extraction refers to the process of slicing, statistically analyzing, and filtering the data corresponding to a continuous time window from the design data set according to the changes in air volume parameters, static pressure parameters, and temperature rise parameters, in order to obtain a set of candidate operating conditions that can characterize different operating states of the target unit.

[0066] Read the air volume parameters, static pressure parameters, and temperature rise parameters from the design data set according to the unit identifier and area identifier, and arrange the air volume parameters, static pressure parameters, and temperature rise parameters corresponding to each time window in chronological order;

[0067] Subsequently, the differences in airflow parameters, static pressure parameters, and temperature rise parameters between the current time window and the next time window are used as the basis for determining the continuity of the operating conditions. Continuity comparisons are performed on adjacent time windows. When the differences in airflow parameters, static pressure parameters, and temperature rise parameters between two adjacent time windows do not exceed their respective operating condition merging thresholds, the two time windows are merged into the same operating condition segment, and the same comparison is performed with subsequent time windows. When any parameter difference exceeds the corresponding operating condition merging threshold, the current operating condition segment is terminated, and the next time window is used as the starting window for the next operating condition segment. Following the above method, segmentation processing is sequentially performed on continuous time windows to obtain a set of operating condition segments; the length of the unified time window is 30 seconds to 120 seconds, preferably 60 seconds; among the operating condition merging thresholds, the merging threshold for air volume parameters is 3% to 5% of the rated air volume, the merging threshold for static pressure parameters is 2% to 4% of the rated static pressure, and the merging threshold for temperature rise parameters is 0.5K to 1.0K; among the merging thresholds, the merging threshold for representative air volume values ​​is 5% to 8% of the rated air volume, the merging threshold for representative static pressure values ​​is 3% to 6% of the rated static pressure, and the merging threshold for representative temperature rise values ​​is 1.0K to 1.5K; the above thresholds are derived from the equipment design specifications corresponding to the target unit type and the statistical calibration results of historical operating samples, and can be adjusted based on actual unit operating data before being put into use.

[0068] The preset coverage requirement is that the cumulative duration corresponding to all representative working condition categories is not less than 85% of the total duration of the target design period; the preset quantity requirement is that the number of representative working condition categories is not less than 5 and not more than 8; when the cumulative coverage is less than 85% or the number of representative working condition categories is less than 5, the merging threshold is appropriately relaxed and the similarity comparison is re-executed until the aforementioned two requirements are met.

[0069] After obtaining the set of operating condition segments, interval statistics are performed on the air volume parameter sequence, static pressure parameter sequence, and temperature rise parameter sequence in each operating condition segment. Specifically: the average value of the air volume value in each window of the air volume parameter sequence is calculated as the representative air volume value corresponding to the operating condition segment; the average value of the static pressure value in each window of the static pressure parameter sequence is calculated as the representative static pressure value corresponding to the operating condition segment; the average value of the temperature rise value in each window of the temperature rise parameter sequence is calculated as the representative temperature rise value corresponding to the operating condition segment; the difference between the load level corresponding to the first window and the load level corresponding to the last window of the operating condition segment is calculated to obtain the load change amplitude corresponding to the operating condition segment; the number of continuous time windows contained in the operating condition segment is multiplied by the uniform time window length to obtain the duration corresponding to the operating condition segment; and the frequency of occurrence of the operating condition segment is obtained according to the number of times that the parameters of the operating condition segment fall within the same parameter range.

[0070] After obtaining the representative values ​​of air volume, static pressure, temperature rise, load change amplitude, duration, and frequency of occurrence, these parameters are combined into a single operating condition feature record in a fixed field order. Each operating condition feature record includes at least the operating condition segment identifier, representative values ​​of air volume, static pressure, temperature rise, load change amplitude, duration, and frequency of occurrence. For operating condition segments with fewer than a preset lower limit of continuous time windows, no operating condition feature record is generated, and the operating condition segment is registered as a discrete segment.

[0071] After obtaining the operating condition characteristic records, a similarity comparison is performed on the representative values ​​of air volume, static pressure, and temperature rise in the operating condition characteristic records. When the differences in the representative values ​​of air volume, static pressure, and temperature rise of two operating condition characteristic records do not exceed the corresponding merging thresholds, the two operating condition characteristic records are merged into the same representative operating condition category, and the representative values ​​of air volume, static pressure, temperature rise, load change amplitude, duration, and frequency of occurrence corresponding to this representative operating condition category are recalculated. When the difference of any parameter exceeds the corresponding merging threshold, they are retained as different representative operating condition categories.

[0072] Subsequently, a retention and screening process is performed on each representative working condition category. Representative working condition categories that meet the preset coverage requirements and the preset number of working condition categories are written into the candidate working condition set, thereby completing the candidate working condition extraction.

[0073] Specifically, the load change amplitude, duration, and frequency of occurrence in the operating condition characteristic records are first read, and each operating condition characteristic record is divided into intervals according to the numerical range of the load change amplitude. Operating condition characteristic records whose load change amplitudes fall within the same numerical range are grouped into the same fluctuation interval, and the corresponding duration and frequency of occurrence are written into the statistical field of the fluctuation interval. Then, the duration within each fluctuation interval is accumulated, and the frequency of occurrence is counted. The load change amplitude range, cumulative duration, and cumulative frequency of occurrence of each operating condition characteristic record within the fluctuation interval are jointly registered as the interval statistical result of the fluctuation interval, thereby obtaining the load fluctuation spectrum.

[0074] The candidate operating condition set refers to a representative set of operating conditions extracted from the design data set that can characterize different load states and changing states during the operation of the target unit. This set serves as the basis for defining subsequent impeller parameter construction, outlet flow field constraint construction, and impeller parameter evaluation. Each candidate operating condition in the candidate operating condition set corresponds to at least three parameters: air volume, static pressure, and temperature rise.

[0075] After obtaining the load fluctuation spectrum, adjacent fluctuation intervals are merged based on similarity. Specifically, the median of the load change amplitude interval, the cumulative duration, and the cumulative frequency of occurrence for each of two adjacent fluctuation intervals are compared. When all three differences do not exceed the corresponding merging threshold, the two fluctuation intervals are merged into the same operating condition category. The median of the merged load change amplitude interval is redefined as the amplitude center of this operating condition category, the cumulative duration is redefined as the duration center of this operating condition category, and the cumulative frequency of occurrence is redefined as the frequency center of this operating condition category. When any difference exceeds the corresponding merging threshold, they are retained as different operating condition categories. The merging process is performed on all fluctuation intervals in the above manner to obtain representative operating condition categories.

[0076] After obtaining the representative operating condition categories, a multi-operating condition design weight generation algorithm is used to calculate the weights for each category. First, the load change amplitude, duration, and frequency of the k-th representative operating condition category are normalized, and its basic fluctuation weight value is calculated. The formula for calculating the basic fluctuation weight value using the multi-operating condition design weight generation algorithm is as follows:

[0077] ,

[0078] in, This represents the basic fluctuation weight value corresponding to the k-th representative working condition category. This represents the load change amplitude corresponding to the k-th representative operating condition category. This represents the duration corresponding to the k-th representative working condition category. This represents the frequency of occurrence of the k-th representative working condition category. Weighting coefficients representing the magnitude of load changes Weighting coefficients representing duration The weight coefficient of the frequency of occurrence weight index, M represents the total number of representative working condition categories, k represents the index of the representative working condition category currently participating in the weight calculation, and i represents the cumulative index when performing normalized summation on all representative working condition categories;

[0079] After obtaining the basic volatility weight value Then, sensitivity correction calculations are performed on the operating condition migration relationships between each representative operating condition category. Specifically, the number of migrations from the k-th representative operating condition category to the j-th representative operating condition category is counted, and the difference in load change amplitude between the k-th and j-th representative operating condition categories is calculated. Subsequently, based on the number of migrations and the difference in load change amplitude, the migration sensitivity correction value corresponding to the k-th representative operating condition category is calculated. The formula for calculating the migration sensitivity correction value of the multi-condition design weight generation algorithm is as follows:

[0080] ,

[0081] in, This represents the migration number from the k-th representative working condition category to the j-th representative working condition category, where j represents the index of other representative working condition categories used for migration relationship calculation with the k-th representative working condition category. This represents the load change amplitude corresponding to the j-th representative operating condition category. This represents the migration-sensitive correction coefficient. This indicates a positive number that prevents the denominator from being zero.

[0082] This correction allows representative operating condition categories that experience frequent transitions and significant changes during operating condition switching to receive higher migration sensitivity correction values. Then, stability gating correction is performed on each representative operating condition category. Specifically, the representative air volume value, static pressure fluctuation value, and temperature rise offset value corresponding to the k-th representative operating condition category are read; when the representative air volume value is lower than the preset low-load air volume threshold, the static pressure fluctuation value is higher than the preset static pressure fluctuation threshold, and the temperature rise offset value is higher than the preset temperature rise offset threshold, the representative operating condition category is determined as a low-load sensitive operating condition, and its stability gating is marked. Set it to 1; otherwise, set its stability gating flag. The low-load sensitive condition is set to 0. Specifically, the air volume representative value corresponding to the k-th representative operating condition category is lower than the preset low-load air volume threshold, and the static pressure fluctuation value is higher than the preset static pressure fluctuation threshold, and the temperature rise offset value is higher than the preset temperature rise offset threshold. The preset low-load air volume threshold is 40% to 60% of the rated air volume, preferably 50%. The preset static pressure fluctuation threshold is 8% to 12% of the rated static pressure, preferably 10%. The preset temperature rise offset threshold is 1.5K to 2.5K, preferably 2.0K.

[0083] The above three thresholds are derived from the statistical results of historical simulation samples of similar high-flow centrifugal fans in the low-load operating range, reflecting the typical triggering conditions for this type of unit to enter the separation risk range.

[0084] Subsequently, based on the stability gating flags Based on the static pressure fluctuation value and temperature rise offset value, the stability gating correction value corresponding to the k-th representative working condition category is calculated. The formula for calculating the stability gating correction value of the multi-working-condition design weight generation algorithm is as follows:

[0085] ,

[0086] in, This represents the stability gating correction value corresponding to the k-th representative operating condition category. This represents the stability gating flag corresponding to the k-th representative operating condition category. This represents the static pressure fluctuation value corresponding to the k-th representative working condition category; This indicates the preset static pressure reference value; This represents the temperature rise offset value corresponding to the k-th representative operating condition category; This indicates the preset temperature rise reference value; Indicates the stability gating correction coefficient;

[0087] After obtaining the basic volatility weight value Migration-sensitive correction value and stability gating correction value Then, multiply the three values ​​together to obtain the comprehensive weight score corresponding to the k-th representative working condition category. ;

[0088] After obtaining the comprehensive weight score corresponding to all representative working condition categories Then, normalization is performed on the scores of each comprehensive weight to obtain the multi-condition design weights corresponding to the k-th representative working condition category. ;

[0089] After the normalization process is completed, the multi-condition design weights corresponding to each representative working condition category are written into the candidate working condition set corresponding to the representative working condition category, so as to complete the determination of the multi-condition design weights corresponding to each candidate working condition in the candidate working condition set, and the multi-condition design weights are used as the working condition weighted input in the subsequent impeller parameter evaluation step.

[0090] The load change amplitude weighting index α is 0.3 to 0.5, the duration weighting index β is 0.2 to 0.4, and the occurrence frequency weighting index γ is 0.2 to 0.4, with the sum of α, β, and γ being 1.0; the migration sensitivity correction coefficient μ is 0.1 to 0.3; the stability gating correction coefficient ν is 0.2 to 0.4; and the preset static pressure reference value... The preset temperature rise reference value is 10% of the rated static pressure. Take 2K; the positive number ε to prevent the denominator from being zero is taken as... ;

[0091] The values ​​of the above coefficients are derived from the simulation calibration results of various typical data center heat load fluctuation scenarios. When using them, they can be adjusted within the given range in combination with the actual load distribution characteristics of the target unit.

[0092] In one embodiment, the multi-condition design weight generation algorithm further includes:

[0093] During the operation of an AI data center, different server racks experience continuously changing thermal load sequences due to training task initiation, computing load switching, and changes in local heat dissipation. The design data set refers to a standardized data set obtained by performing time alignment, duplicate data processing, outlier removal, and unit unification on the impeller optimization design input data. This standardized data set serves as a common input for candidate operating condition extraction, impeller parameter construction, and outlet flow field constraint construction. A unified time window correspondence and unified field caliber have been established for each data item in the design data set. To ensure that subsequent impeller optimization design can cover high-load continuous operation, rapid load increase, load decrease, and medium-to-low load dwell conditions, the load change amplitude, duration, and frequency of occurrence in the operating condition characteristic records are first jointly read. Operating condition characteristic records with similar load change amplitudes, durations, and frequencies are grouped into the same representative operating condition category to obtain the load fluctuation spectrum corresponding to the cabinet thermal load change process. Then, the operating condition importance of each representative operating condition category in the load fluctuation spectrum is calculated so that the representative operating condition categories with large load change amplitudes, long durations, and high frequencies receive higher basic design weights.

[0094] After calculating the basic design weights, the switching relationships between representative operating condition categories are read and compared. When a representative operating condition category repeatedly appears in the switching interval between high and medium loads, or repeatedly appears in the switching interval between medium and low loads, this representative operating condition category is identified as a sensitive category for operating condition migration, and the basic design weight corresponding to this representative operating condition category is adjusted for migration. When a representative operating condition category appears infrequently but is in a frequent switching link, and the load change difference between it and adjacent operating condition categories is large, its design weight is also increased. Through the above processing, transitional operating conditions that are easily underestimated under single-frequency statistics are included in the subsequent impeller design weight system, so that the candidate operating condition set no longer only covers stable residence operating conditions, but also covers switching operating conditions that have a significant impact on impeller stability.

[0095] After completing the migration correction process, low-load sensitivity identification is performed on each representative operating condition category. For representative operating condition categories exhibiting a decrease in representative airflow value, increased static pressure fluctuation, and a continuous increase in temperature rise deviation, these are identified as low-load sensitive operating conditions, and the design weight corresponding to these representative operating condition categories is subject to gating enhancement. This process ensures that operating condition categories related to blade separation, enhanced recirculation, and wide-range instability risks have a higher weighting in subsequent impeller parameter evaluations. This allows the subsequent evaluation process to prioritize excluding impeller parameter combinations that perform well only under rated operating conditions but lack stability under medium- and low-load fluctuation conditions.

[0096] After completing the basic design weight calculation, migration correction processing, and low-load sensitive gating processing, the multi-condition design weights corresponding to each candidate operating condition are obtained, and these multi-condition design weights are written into the candidate operating condition set. Subsequently, when performing impeller parameter evaluation, the aerodynamic performance evaluation results, outlet flow field uniformity evaluation results, low-load separation risk evaluation results, and blade frequency noise proxy evaluation results are all weighted according to the multi-condition design weights. This ensures that the impeller optimization design process is not carried out around a single design point, but rather around a multi-condition chain corresponding to the actual cabinet heat load fluctuation process. Through the above method, the multi-condition design weight generation algorithm transforms the cabinet heat load fluctuation spectrum into the operating condition weighting basis for the impeller optimization design stage, so that the obtained target impeller parameter results simultaneously take into account the aerodynamic performance under high air volume operating conditions and the stability under wide operating condition conditions.

[0097] The impeller parameters are constructed from the candidate operating condition set to obtain an optimizable impeller parameter set;

[0098] The impeller parameter construction refers to the process of transforming and expanding the applicable boundaries and geometric parameters of the impeller based on the operating parameters, reference impeller parameters and installation boundary information corresponding to the candidate operating condition set, in order to obtain an optimizable impeller parameter set that can be used for subsequent evaluation and solution.

[0099] The impeller design boundary transformation specifically involves: first, reading the airflow parameters, static pressure parameters, and temperature rise parameters corresponding to each candidate operating condition in the candidate operating condition set; calculating the average inlet velocity and average outlet velocity of the impeller based on the airflow parameters, using them as the design boundary input for the blade installation angle parameters; determining the impeller design static pressure rise requirement based on the static pressure parameters, and determining the feasible range of the blade wrap angle parameters and blade number parameters in conjunction with the impeller outer diameter and speed range registered in the installation boundary information; determining the airflow density correction coefficient based on the temperature rise parameters, and writing this correction coefficient into the density boundary field of the impeller geometric parameter expansion. For any candidate operating condition where the boundary parameters obtained from the above transformation contradict the allowable range registered in the installation boundary information, the allowable range in the installation boundary information is taken as the standard, and the contradictory boundary parameters are truncated, with a truncation mark written into the impeller design boundary parameter record. After completing the above transformation, the blade installation angle boundary, blade wrap angle boundary, blade number boundary, and density correction coefficient corresponding to each candidate operating condition are written into the impeller design boundary parameters to obtain the impeller design boundary parameter set.

[0100] The impeller geometry parameter expansion specifically involves: first reading the baseline impeller parameters from the design data set, and then extracting the baseline inlet installation angle. Reference outlet installation angle Reference blade wrap angle Reference number of blades The reference values ​​for the reference blade thickness control points t1 to t5, and the reference leading edge sweep. and baseline trailing edge unloading amount Next, based on the blade mounting angle boundary, blade wrap angle boundary, and blade number boundary corresponding to each candidate operating condition in the impeller design boundary parameter set, candidate value sequences for each parameter are generated within the allowable range of each boundary at a preset step size. Subsequently, a full combination expansion is performed on the candidate value sequences of blade mounting angle parameters, blade wrap angle parameters, blade number parameters, split blade length ratio parameters, blade thickness distribution parameters, leading edge sweep, and trailing edge unloading. For each geometric parameter combination, it is checked whether it satisfies all geometric constraints registered in the mounting boundary information. For geometric parameter combinations that satisfy all geometric constraints, they are added to the optimizable impeller parameter set. For geometric parameter combinations that violate any constraint, they are removed from the expansion results and not added to the optimizable impeller parameter set. After completing the above expansion, an optimizable impeller parameter set is obtained that can be used for subsequent evaluation and solution.

[0101] The outlet flow field constraint is constructed by performing an outlet flow field constraint on the candidate working condition set to obtain the outlet flow field constraint set;

[0102] The outlet flow field constraint construction refers to the process of performing correlation analysis, outlet partition mapping, and constraint quantification on the sensitivity of the terminal region based on the terminal temperature rise deviation data, regional pressure difference fluctuation data, and reflux characterization data corresponding to the candidate operating condition set, in order to obtain the outlet flow field constraint set corresponding to each candidate operating condition.

[0103] Specifically, firstly, the terminal temperature rise deviation data, regional pressure difference fluctuation data, and recirculation characterization data corresponding to the candidate operating condition set in the design data set are read; then, sensitivity correlation analysis is performed on the terminal temperature rise deviation data, the regional pressure difference fluctuation data, and the recirculation characterization data to obtain a hot spot sensitivity matrix; subsequently, the hot spot sensitivity matrix is ​​mapped to the outlet partition and constrained to obtain an outlet flow field constraint set containing non-uniform penalty constraints for each outlet partition.

[0104] In one embodiment, the hot spot sensitivity matrix is ​​subjected to outlet partition mapping and constraint quantization, specifically including: matching the sensitivity values ​​of each region in the hot spot sensitivity matrix with the corresponding partitions of the wind turbine outlet to obtain partition sensitivity records; calculating the penalty values ​​of the partition sensitivity records using an outlet non-uniform penalty constraint generation algorithm based on hot spot sensitivity inverse mapping, to obtain the non-uniform penalty constraints corresponding to each outlet partition; and performing partition combination encapsulation on the non-uniform penalty constraints corresponding to each outlet partition to complete the construction of an outlet flow field constraint set containing the non-uniform penalty constraints of each outlet partition.

[0105] The impeller parameter evaluation is performed on the optimizable impeller parameter set and the outlet flow field constraint set to obtain the comprehensive impeller evaluation result;

[0106] Specifically, first, each impeller parameter combination in the optimizable impeller parameter set is read item by item according to the impeller parameter combination identifier, and then each candidate operating condition parameter in the candidate operating condition set is read item by item according to the candidate operating condition identifier; then, the multi-operating condition design weights corresponding to each candidate operating condition and the non-uniform penalty constraints of each outlet partition corresponding to each candidate operating condition in the outlet flow field constraint set are read item by item according to the candidate operating condition identifier.

[0107] The outlet flow field constraint set refers to the constraint set obtained after sensitivity correlation analysis, outlet partition mapping, and constraint quantification based on the terminal temperature rise deviation data, regional pressure difference fluctuation data, and recirculation characterization data corresponding to the candidate operating condition set. It is used to limit the allowable deviation range of the velocity distribution in each outlet partition during the impeller parameter evaluation process. Each constraint item in the outlet flow field constraint set corresponds at least to a candidate operating condition identifier, an outlet partition identifier, and a non-uniformity penalty constraint value.

[0108] Subsequently, an association registration is performed on the current impeller parameter combination, the current candidate operating condition parameters, the multi-condition design weights corresponding to the current candidate operating condition, and the non-uniform penalty constraints of each outlet zone corresponding to the current candidate operating condition. Specifically, the association registration involves: first, using the candidate operating condition identifier as the first matching key, retrieving data items with the same candidate operating condition identifier from the candidate operating condition set, the multi-condition design weight set, and the outlet flow field constraint set; then, using the impeller parameter combination identifier as the second matching key, combining the current impeller parameter combination with the retrieved candidate operating condition parameters, multi-condition design weights, and non-uniform penalty constraints of each outlet zone into the same evaluation item; subsequently, the impeller parameter combination identifier, candidate operating condition identifier, air volume parameter, static pressure parameter, temperature rise parameter, multi-condition design weights, and non-uniform penalty constraints of each outlet zone are written into the evaluation item record in a unified field order. If any candidate operating condition identifier lacks candidate operating condition parameters, multi-condition design weights, or outlet flow field constraints, the current association registration is terminated, and the evaluation item is recorded as an invalid evaluation item.

[0109] After obtaining valid evaluation items, aerodynamic simulation is performed on each evaluation item in sequence. Specifically: First, the blade installation angle parameter, blade wrap angle parameter, number of blades parameter, split blade length ratio parameter, blade thickness distribution parameter, leading edge sweep and trailing edge unloading amount in the impeller parameter combination corresponding to the current evaluation item are written into the impeller geometry model one by one; then, the air volume parameter, static pressure parameter and temperature rise parameter corresponding to the current evaluation item are written into the operating condition boundary conditions one by one, and the current candidate operating condition parameters are made to correspond one-to-one with the current impeller geometry model; then, the flow field is solved for the current evaluation item, and the total pressure value of the impeller inlet section, the total pressure value of the impeller outlet section, the flow velocity value of each zone at the outlet, the coverage ratio of the low-speed zone of the impeller passage, the coverage ratio of the counter-flow zone of the impeller passage, and the pressure pulsation value are read from the solution results; finally, the above solution results are written into the operating condition aerodynamic performance record in the order of evaluation item recording. For evaluation items where the flow field solution fails to meet the preset convergence condition, or where any one of the following is missing: inlet total pressure, outlet total pressure, outlet velocity values ​​in each zone, or pressure pulsation values, it is recorded as a simulation failure evaluation item, and the corresponding impeller parameter combination is excluded from the subsequent comprehensive evaluation.

[0110] The aerodynamic simulation was performed using a steady RANS solver based on the finite volume method, and the turbulence model adopted was the SSTk-ω model. The computational domain covered the extended section of the impeller inlet, the impeller flow channel, and the extended section of the volute outlet. The mesh adopted a hybrid mesh dominated by hexahedrons, and the height of the first layer of mesh on the blade wall was calculated according to... Control measures include a minimum of 800,000 grid cells per impeller channel and a minimum of 600,000 grid cells per volute region. Boundary conditions include a given total pressure and temperature at the inlet, a given static pressure at the outlet, and a non-slip adiabatic wall. The discretization scheme uses a second-order upwind scheme for convection and the SIMPLE algorithm for pressure-velocity coupling. The convergence criterion is that the normalized residuals of each equation decrease to a minimum. The following conditions apply, and the inlet and outlet mass flow rate deviation is less than 0.5%. When the solution reaches the convergence criterion, the following parameters are extracted from the solution: impeller inlet section mass-weighted average total pressure, impeller outlet section mass-weighted average total pressure, outlet annular area-weighted average radial velocity, impeller passage low-velocity zone coverage ratio, impeller passage counter-current zone coverage ratio, and pressure time history at monitoring points. The pressure pulsation value is taken as the impeller passing frequency. The root mean square value of the spectral amplitude at that location.

[0111] After obtaining the aerodynamic performance records for the operating conditions, static pressure efficiency calculation, outlet flow field uniformity calculation, low-load separation risk calculation, and blade frequency noise proxy value calculation are performed for each aerodynamic performance record. Specifically: First, the static pressure efficiency value corresponding to the current impeller parameter combination is calculated based on the air volume parameters, impeller inlet total pressure value, impeller outlet total pressure value, and power value under the current candidate operating condition; then, the average flow velocity value is obtained for all outlet zone flow velocity values ​​corresponding to the current evaluation item, and the absolute value of the deviation between the flow velocity value of each outlet zone and the average flow velocity value is calculated one by one. The sum of all absolute deviation values ​​is divided by the total number of outlet zones to obtain the outlet flow field uniformity value corresponding to the current impeller parameter combination; subsequently, the low-speed zone coverage ratio, the counter-flow zone coverage ratio, and the pressure pulsation value of the impeller passage are read, and the low-speed zone coverage ratio, the counter-flow zone coverage ratio, and the pressure pulsation value are converted into a low-load separation risk value according to a preset risk mapping rule; then, the pressure pulsation value, blade number parameters, and rotation operating condition parameters are converted into a blade frequency noise proxy value.

[0112] The calculation method for the low-load separation risk value is as follows: first, read the coverage ratio of the low-speed zone of the blade passage corresponding to the current evaluation item. Coverage ratio of the backflow zone in the leaf aisle and pressure pulsation value The low-velocity zone of the blade passage is defined as the area where the relative velocity within the blade passage is less than 60% of the average relative velocity within the blade passage, and the counter-current zone of the blade passage is defined as the area where the angle between the relative velocity vector and the mainstream direction of the blade passage is greater than 90°; then according to the formula... Converted to low-load separation risk value, The risk weight representing the proportion of low-speed zone coverage in the leaf passage. Risk weights representing the coverage ratio of the duct countercurrent zone. The risk weights represent the normalized values ​​of pressure pulsations. The sum is taken as 1.0. Δpref is taken as 5% of the rated static pressure.

[0113] The method for calculating the blade frequency noise proxy value is as follows: first, read the pressure pulsation value Δp, the number of blades Z, and the rated speed n corresponding to the current evaluation item, and then calculate the blade frequency. ;

[0114] Then follow the formula Converted to blade frequency noise proxy value, where Pick , Pick ;

[0115] Wherein, n represents the impeller speed corresponding to the current candidate operating condition, in rpm, which is taken from the speed record item in the rotational operating condition parameters;

[0116] Where δ represents a positive number to prevent the logarithmic argument from being zero, and is taken as 1 × 10⁻³;

[0117] This blade frequency noise proxy value is only used for relative comparison between different impeller parameter combinations and is not used as an absolute noise value.

[0118] Finally, the static pressure efficiency value, outlet flow field uniformity value, low load separation risk value, and blade frequency noise proxy value are written into the sub-condition evaluation record. For any sub-condition evaluation record where any evaluation value cannot be calculated or exceeds the preset effective range, it is recorded as an abnormal evaluation record and is not included in the subsequent weighted summary.

[0119] After obtaining the sub-condition evaluation records, all sub-condition evaluation records belonging to the same impeller parameter combination are aggregated according to the candidate condition identifier, and weighted summation is performed according to the multi-condition design weight corresponding to each candidate condition. The aggregation is specifically as follows: first, using the impeller parameter combination identifier as the grouping key, the sub-condition evaluation records belonging to the same impeller parameter combination are grouped into the same parameter combination group; then, all sub-condition evaluation records in the same parameter combination group are sorted according to the candidate condition identifier; subsequently, the sorted sub-condition evaluation records are read one by one, and each is multiplied by the multi-condition design weight of the corresponding candidate condition. The weighted summation process is as follows: The static pressure efficiency value corresponding to each sub-condition evaluation record is multiplied by its multi-condition design weight and summed to obtain the weighted static pressure efficiency value; the outlet flow field uniformity value is multiplied by its multi-condition design weight and summed to obtain the weighted flow field uniformity value; the low-load separation risk value is multiplied by its multi-condition design weight and summed to obtain the weighted separation risk value; the blade frequency noise proxy value is multiplied by its multi-condition design weight and summed to obtain the weighted noise proxy value; finally, the weighted static pressure efficiency value, weighted flow field uniformity value, weighted separation risk value, and weighted noise proxy value are written into the weighted evaluation record. For impeller parameter combinations that lack effective sub-condition evaluation records under any high-weight candidate condition, they are recorded as incomplete evaluation parameter combinations and excluded from subsequent comprehensive ranking.

[0120] After obtaining the weighted evaluation records, an outlet flow field constraint compliance evaluation is performed on each impeller parameter combination. Specifically, the flow velocity values ​​of each outlet zone corresponding to the current impeller parameter combination under each candidate operating condition are first read; then, the non-uniform penalty constraints of each outlet zone are read item by item according to the candidate operating condition identifier and the outlet zone identifier; subsequently, the flow velocity deviation of the same outlet zone under the same candidate operating condition is compared with the non-uniform penalty constraint corresponding to that outlet zone. When the flow velocity deviation does not exceed the corresponding non-uniform penalty constraint, the outlet zone is recorded as a constraint-passing zone; when the flow velocity deviation exceeds the corresponding non-uniform penalty constraint, the outlet zone is recorded as a constraint-failing zone, and the flow velocity deviation of the outlet zone is subtracted from the non-uniform penalty constraint to obtain the excess value corresponding to that outlet zone. After completing the comparison of all outlet zones, the number of constraint-failing zones corresponding to the current impeller parameter combination is counted, and the excess values ​​of all constraint-failing zones are summed to obtain the total excess value. If the number of constraint failure zones is zero, the weighted evaluation record corresponding to the current impeller parameter combination is directly retained; if the number of constraint failure zones is greater than zero, the weighted flow field uniformity value is reduced by the penalty value corresponding to the sum of the out-of-limit values, and the comprehensive evaluation level is downgraded by one or more levels to obtain a corrected evaluation record; if the number of constraint failure zones exceeds the preset upper limit, the current impeller parameter combination is recorded as a constraint mismatch parameter combination and excluded from subsequent optimization processing.

[0121] After obtaining the corrected evaluation records, a comprehensive ranking is performed on all impeller parameter combinations. Specifically: first, parameter combinations with constraint mismatch and incomplete evaluations are removed from the candidate ranking set; then, the remaining impeller parameter combinations are grouped according to their comprehensive evaluation level; subsequently, within the same comprehensive evaluation level, the weighted static pressure efficiency value is compared first, and the impeller parameter combination with the larger value is ranked first; when the weighted static pressure efficiency values ​​are the same, the weighted flow field uniformity value is compared, and the impeller parameter combination with the larger value is ranked first; when the weighted flow field uniformity values ​​are still the same, the weighted separation risk value is compared, and the impeller parameter combination with the smaller value is ranked first; when the weighted separation risk value is still the same, the weighted noise proxy value is compared, and the impeller parameter combination with the smaller value is ranked first; if the above four comparisons still cannot distinguish the order, the impeller parameter combination with fewer constraint failure zones is retained first; if the number of constraint failure zones is the same, the impeller parameter combination with the smaller total over-limit value is retained first. After sorting, the impeller parameter combinations that are within the preset retention range are registered as the preferred parameter combinations, and the comprehensive evaluation results of the impeller are obtained.

[0122] The comprehensive evaluation result of the impeller refers to the comprehensive evaluation result obtained after performing aerodynamic simulation, sub-condition evaluation, multi-condition weighted summarization, and outlet flow field constraint compliance correction on each combination of impeller parameters in the optimizable impeller parameter set under each candidate operating condition. This result is used as the input for the optimization solution. The comprehensive evaluation result of the impeller includes at least the comprehensive evaluation level, weighted static pressure efficiency value, weighted flow field uniformity value, weighted separation risk value, weighted noise proxy value, number of constraint-missing zones, and the sum of out-of-limit values.

[0123] The comprehensive evaluation results of the impeller are optimized and solved to obtain the target impeller parameters.

[0124] The optimization solution process refers to the process of prioritizing, verifying stability under wide operating conditions, verifying compliance with installation boundaries, and iteratively solving the impeller parameter combination based on the comprehensive evaluation results of the impeller, in order to obtain the target impeller parameter results that meet the evaluation requirements and constraints.

[0125] Specifically, the impeller parameter combinations in the comprehensive evaluation results of the impeller are read one by one, and the impeller parameter combination identifier, comprehensive evaluation level, weighted static pressure efficiency value, weighted flow field uniformity value, weighted separation risk value, weighted noise proxy value, number of constraint failure zones and total over-limit value are extracted sequentially from each impeller parameter combination.

[0126] Subsequently, based on the comprehensive evaluation level, impeller parameter combinations with the same comprehensive evaluation level are grouped into the same evaluation level group. Then, within each evaluation level group, the impeller parameter combinations are arranged in ascending order by their identifiers, and the weighted static pressure efficiency, weighted flow field uniformity, weighted separation risk, weighted noise proxy, number of constraint-failed zones, and sum of out-of-limit values ​​are compared sequentially between adjacent impeller parameter combinations within the same evaluation level group. When different weighted static pressure efficiency values ​​occur, the impeller parameter combination with the larger value is ranked first; when the same weighted static pressure efficiency values ​​occur... Continue comparing the weighted flow field uniformity values, prioritizing impeller parameter combinations with larger values. If the weighted flow field uniformity values ​​remain the same, continue comparing the weighted separation risk values, prioritizing impeller parameter combinations with smaller values. If the weighted separation risk values ​​remain the same, continue comparing the weighted noise surrogate values, prioritizing impeller parameter combinations with smaller values. If the above four comparisons still cannot distinguish the order, then compare the number of constraint-failed zones and the total number of out-of-limit values, prioritizing impeller parameter combinations with fewer constraint-failed zones and a smaller total number of out-of-limit values. After completing the arrangement within the same evaluation level group, connect the evaluation level groups sequentially from high to low according to the comprehensive evaluation level to obtain the collaborative optimization candidate sequence.

[0127] After obtaining the collaborative optimization candidate sequence, the wide operating condition stability check and installation boundary compliance check are performed on the collaborative optimization candidate sequence in sequence. Specifically, the process is as follows: First, the weighted separation risk value, weighted flow field uniformity value, and weighted static pressure efficiency value corresponding to each impeller parameter combination are read sequentially according to the order of arrangement in the collaborative optimization candidate sequence. Then, the weighted separation risk value is compared with the preset stability upper limit item by item, the weighted flow field uniformity value is compared with the preset uniformity lower limit item by item, and the weighted static pressure efficiency value is compared with the preset efficiency lower limit item by item. When the weighted separation risk value is not higher than the preset stability upper limit, the weighted flow field uniformity value is not lower than the preset uniformity lower limit, and the weighted static pressure efficiency value is not lower than the preset efficiency lower limit, the current impeller parameter combination is recorded as a wide-condition stability pass parameter combination. When any indicator does not meet the corresponding threshold condition, the current impeller parameter combination is recorded as a wide-condition stability fail parameter combination, and the impeller parameter combination is excluded from the subsequent feasible parameter screening. At the same time, the specific indicator items that cause the verification failure (separation risk value exceeding the limit, flow field uniformity value insufficient, or static pressure efficiency value insufficient) are written into the verification failure reason field of the impeller parameter combination for subsequent analysis of the failure reasons.

[0128] The preset stability upper limit corresponds to the weighted separation risk value, which is between 0.30 and 0.40; the preset uniformity lower limit corresponds to the weighted flow field uniformity value, which is between 0.75 and 0.85; and the preset efficiency lower limit corresponds to the weighted static pressure efficiency value, which is between 0.72 and 0.78.

[0129] Subsequently, for wide-condition stability testing, the blade installation angle, blade wrap angle, number of blades, splitter blade length ratio, blade thickness distribution, leading-edge sweep, and trailing-edge unloading parameters are continuously read and compared with the installation boundary information in the design data set. When all parameter values ​​are within the allowable range of the corresponding installation boundary, the current impeller parameter combination is recorded as an installation boundary pass parameter combination. When any parameter value exceeds the allowable range of the corresponding installation boundary, the current impeller parameter combination is recorded as a wide-condition stability fail parameter combination and excluded from subsequent feasible parameter screening. After completing the above two checks, the impeller parameter combinations that are simultaneously recorded as wide-condition stability pass parameter combinations and installation boundary pass parameter combinations are written into the feasible parameter sequence in their original order.

[0130] After obtaining the feasible parameter sequence, an iterative solution is performed on the feasible parameter sequence using an aerodynamic thermal field noise collaborative optimization algorithm. Specifically, all impeller parameter combinations are first renumbered according to the order in the feasible parameter sequence, and the first... impeller parameter combination in the first The q-th parameter value in the round of iteration is denoted as Where q corresponds to the blade installation angle parameter, blade wrap angle parameter, splitter blade length ratio parameter, leading edge sweep, trailing edge unloading, and other geometric parameters involved in the optimization; then, each parameter is read sequentially. Weighted hydrostatic efficiency values ​​corresponding to the combination of impeller parameters in the wheel Weighted flow field homogeneity value Weighted Separation Risk Value Weighted noise proxy value Constraints on the number of partitions not passed and the sum of the excess values The constraint pressure value and the collaborative optimization objective value corresponding to each impeller parameter combination are calculated sequentially. The formula for calculating the constraint pressure value is as follows:

[0131] ,

[0132] in, Indicates the first Wheel of Life The constraint pressure value corresponding to the impeller parameter combination; Indicates the first Wheel of Life The constraints corresponding to the impeller parameter combinations do not depend on the number of partitions; Indicates the first Wheel of Life The sum of the out-of-limit values ​​corresponding to the parameter combinations of the impeller; This indicates that the constraint does not pass the reference value for the number of partitions; This represents the reference value for the sum of the out-of-limit values; Indicates the pressure coefficient of the number of partitions; Indicates the pressure coefficient exceeding the limit; This indicates a positive number that prevents the denominator from being zero.

[0133] After obtaining the constraint pressure value, the collaborative optimization objective value corresponding to each impeller parameter combination is calculated. The formula for calculating the collaborative optimization objective value is as follows:

[0134] ,

[0135] in, Indicates the first Wheel of Life The collaborative optimization objective value corresponding to the combination of impeller parameters; Indicates the first Wheel of Life Weighted hydrostatic efficiency value corresponding to the parameter combination of the impeller; Indicates the first Wheel of Life Weighted flow field uniformity values ​​corresponding to the combination of impeller parameters; Indicates the first Wheel of Life Weighted separation risk value corresponding to the impeller parameter combination; Indicates the first Wheel of Life Weighted noise proxy value corresponding to the parameter combination of the impeller; Indicates the static pressure efficiency gain index; Indicates the gain index of flow field uniformity; This represents the risk penalty coefficient for separation. Indicates the noise penalty factor; This represents the constraint pressure penalty coefficient.

[0136] After obtaining the collaborative optimization objective values ​​corresponding to all impeller parameter combinations, the first... The collaborative optimization objective values ​​of all impeller parameter combinations in the current impeller are compared. The impeller parameter combination with the largest collaborative optimization objective value is determined as the optimal impeller parameter combination for the current impeller, and its corresponding collaborative optimization objective value is denoted as [value]. The corresponding constraint pressure value is denoted as .

[0137] Subsequently, using the optimal impeller parameter combination of the current impeller as the baseline parameter combination, parameter perturbations are sequentially performed on its first to Qth optimizable parameter items; the parameter perturbation specifically involves: first reading the perturbation step size corresponding to the qth optimizable parameter item in the current impeller. Then, the q-th parameter value in the current optimal impeller parameter combination is added to and subtracted from the disturbance step size to obtain two parameter disturbance values. Two new impeller parameter combinations are then generated while keeping the remaining parameter values ​​unchanged. For these two new impeller parameter combinations, it is first checked whether each parameter value is within the allowable range registered in the installation boundary information. For impeller parameter combinations where all parameter values ​​meet the installation boundary requirements, they are sequentially written into the q-th parameter combination. The set of candidate impeller parameter combinations is defined, where the parameter values ​​not involved in the current parameter item disturbance remain completely consistent with the current optimal impeller parameter combination. For any impeller parameter combination where any parameter value exceeds the allowable range of the installation boundary, this impeller parameter combination is marked as a boundary violation parameter combination and is not included in the first iteration. Round the candidate set, and shrink the perturbation step size of the current parameter item according to the preset shrinkage ratio and regenerate the perturbation value.

[0138] Then, apply the same forward and reverse perturbations to the next optimizable parameter term, continuing this process until all optimizable parameter terms have been perturbed. After perturbing all parameter terms, the result is the [number of parameters]. Candidate impeller parameter combinations. The parameter step size update formula is:

[0139] ,

[0140] in, Indicates the first The perturbation step size corresponding to the qth optimizable parameter in the round; Indicates the first The perturbation step size corresponding to the qth optimizable parameter in the round; Indicates the first The collaborative optimization objective value corresponding to the optimal impeller parameter combination of the current impeller in the wheel; Indicates the first The collaborative optimization objective value corresponding to the optimal impeller parameter combination of the previous impeller in the wheel; Indicates the first The constraint pressure value corresponding to the optimal impeller parameter combination of the current impeller in the wheel; Indicates the target value contraction coefficient; This represents the constraint pressure amplification coefficient. Using the updated parameter step sizes, the [parameter name] is [value] for the [parameter name]. The candidate impeller parameter combinations are re-evaluated, verified, and target value calculated to proceed to the next iteration.

[0141] After completing two consecutive iterations, a convergence determination is performed on the optimal impeller parameter combination for the current iteration. The convergence determination formula is as follows:

[0142] ,

[0143] in, Indicates the first The convergence criterion value of the wheel; Indicates the first The constraint pressure value corresponding to the optimal impeller parameter combination of the previous impeller in the wheel;

[0144] when When the value is less than or equal to the preset convergence threshold, the optimization solution process is terminated, and the optimal impeller parameter combination of the current impeller is determined as the target impeller parameter result.

[0145] when When the value exceeds the preset convergence threshold, the optimal impeller parameter combination of the current round is used as the solution benchmark for the next round, and parameter perturbation, impeller parameter evaluation, verification comparison and convergence determination are continued.

[0146] The preset convergence threshold is between 0.5% and 1.0%; the preset upper limit of the number of constraint failure partitions is 25% of the total number of exit partitions, which is 2 when the total number of exit partitions is 8 and 3 when the total number of exit partitions is 12.

[0147] The pressure coefficient of the number of partitions The pressure coefficient exceeding the limit is taken as 0.2 to 0.4. Take a value between 0.3 and 0.5; the constraint does not pass the reference value for the number of partitions. Take 10% to 15% of the total number of export zones, and the sum of the aforementioned exceedance values ​​is a reference value. Take 8% of the rated outlet flow rate.

[0148] The static pressure efficiency gain index The flow field uniformity gain exponent is set to 1.0 to 1.5. The value is taken as 0.5 to 1.0; the separation risk penalty coefficient. The noise penalty coefficient is set to between 0.3 and 0.6. The constraint pressure penalty coefficient is set to between 0.2 and 0.4. Take a value between 0.5 and 1.0.

[0149] The target value shrinkage coefficient The constraint pressure amplification coefficient is taken as 0.3 to 0.6. Take values ​​from 0.1 to 0.3; the initial perturbation step size for each optimizable parameter is 5% to 10% of the current value, and the corresponding reduction ratio of the perturbation step size in a single iteration is 0.5 to 0.7;

[0150] The values ​​of the above coefficients are derived from the simulation calibration results of multiple impeller parameter optimization scenarios. The relative proportions between the coefficients reflect the engineering priority trade-off between static pressure efficiency and flow field uniformity.

[0151] For cases where the feasible parameter sequence is empty, the impeller parameter combinations are first reread in the order of the candidate sequence for collaborative optimization, and then the wide-condition stability check and installation boundary compliance check are re-executed. If no feasible parameter sequence is obtained after re-checking, the impeller parameter combinations that are ranked high in the candidate sequence for collaborative optimization, have fewer constraint failure partitions, and have a smaller total sum of out-of-limit values ​​are read. The perturbation step size of each optimizable parameter item is reduced according to a preset shrinkage ratio, and a new candidate impeller parameter combination is generated and iterative solution is continued. After the above processing, the target impeller parameter results that meet the wide-condition stability requirements and installation boundary requirements are obtained. The target impeller parameter results refer to the impeller parameter results determined after priority sorting, wide-condition stability check, installation boundary compliance check, and iterative solution based on the impeller comprehensive evaluation results. These results are used as the design output of the large-flow centrifugal fan impeller corresponding to the target unit. The parameter values ​​in the target impeller parameter results meet the installation boundary requirements and correspond to the impeller design values ​​in subsequent engineering implementation.

[0152] In one embodiment, the aerodynamic thermal field noise co-optimization solution algorithm further includes:

[0153] In one example scenario, multiple sets of comprehensive impeller evaluation results have been obtained for the same target unit. These comprehensive evaluation results include at least the comprehensive evaluation level, weighted static pressure efficiency value, weighted flow field uniformity value, weighted separation risk value, weighted noise proxy value, number of constraint-failed zones, and total number of out-of-limit values ​​corresponding to the first, second, and third impeller parameter combinations, respectively. First, all impeller parameter combinations are categorized according to their comprehensive evaluation level, with combinations having the same comprehensive evaluation level grouped together. Then, within each evaluation level group, the weighted static pressure efficiency value, weighted flow field uniformity value, weighted separation risk value, and weighted noise proxy value are compared sequentially. Impeller parameter combinations with higher comprehensive evaluation levels and larger weighted static pressure efficiency values, larger weighted flow field uniformity values, smaller weighted separation risk values, and smaller weighted noise proxy values ​​are ranked first, thus obtaining a collaborative optimization candidate sequence. For example, if the first impeller parameter combination and the second impeller parameter combination are in the same comprehensive evaluation level group, and the weighted static pressure efficiency value of the first impeller parameter combination is higher than that of the second impeller parameter combination, then the first impeller parameter combination is ranked before the second impeller parameter combination; if the weighted static pressure efficiency values ​​are the same, then the weighted flow field uniformity values ​​are compared, and the impeller parameter combination with the higher weighted flow field uniformity value is ranked first.

[0154] After obtaining the candidate sequence for collaborative optimization, a wide-condition stability check is performed on each impeller parameter combination in the candidate sequence. Specifically, the weighted separation risk value, weighted flow field uniformity value, and weighted static pressure efficiency value corresponding to the current impeller parameter combination are read first. Then, the weighted separation risk value is compared with the preset stability upper limit item by item, the weighted flow field uniformity value is compared with the preset uniformity lower limit item by item, and the weighted static pressure efficiency value is compared with the preset efficiency lower limit item by item. When the weighted separation risk value is not higher than the preset stability upper limit, the weighted flow field uniformity value is not lower than the preset uniformity lower limit, and the weighted static pressure efficiency value is not lower than the preset efficiency lower limit, the impeller parameter combination is determined as a wide-condition stability pass parameter combination. When any item does not meet the corresponding threshold condition, the impeller parameter combination is determined as an installation boundary fail parameter combination and is excluded from the subsequent feasible parameter screening. For example, if the weighted separation risk value of the first impeller parameter combination is lower than the preset stability upper limit, and the weighted flow field uniformity value and the weighted static pressure efficiency value both meet the corresponding threshold requirements, then the first impeller parameter combination is retained; if the weighted static pressure efficiency value of the second impeller parameter combination is higher than the preset stability upper limit, then the second impeller parameter combination is excluded from the subsequent processing.

[0155] After completing the wide-condition stability verification, the impeller parameter combinations that passed the wide-condition stability verification are further subjected to installation boundary compliance verification. Specifically, the blade installation angle parameter, blade wrap angle parameter, number of blades parameter, split blade length ratio parameter, blade thickness distribution parameter, leading edge sweep, and trailing edge unloading amount in the current impeller parameter combination are read item by item and compared with the installation boundary information registered in the design data set. When all parameter values ​​are within the allowable range of the corresponding installation boundary, the current impeller parameter combination is determined to be an installation boundary pass parameter combination; when any parameter value exceeds the allowable range of the corresponding installation boundary, the current impeller parameter combination is determined to be an installation boundary fail parameter combination and is excluded from the subsequent feasible parameter screening. For example, if the blade installation angle parameter and leading edge sweep of the first impeller parameter combination both fall within the allowable range of the installation boundary, the first impeller parameter combination is retained; if the split blade length ratio parameter of the third impeller parameter combination exceeds the allowable range of the installation boundary, the third impeller parameter combination is excluded. After completing the above two verifications, the impeller parameter combination that simultaneously passes the wide operating condition stability verification and the installation boundary compliance verification will be written into the feasible parameter sequence.

[0156] After obtaining the feasible parameter sequence, iterative solution is performed on the feasible parameter sequence. Specifically, the first impeller parameter combination in the feasible parameter sequence is read and determined as the current impeller reference parameter combination; then, parameter perturbation is performed on each of the blade installation angle parameter, blade wrap angle parameter, splitter blade length ratio parameter, leading edge sweep, and trailing edge unloading in the current impeller reference parameter combination. When performing parameter perturbation, first select a parameter to be adjusted, then increase and decrease the perturbation step size by one step size respectively based on the current value of the parameter to be adjusted, while keeping the other parameters unchanged, thus obtaining two new parameter adjustment combinations. Subsequently, aerodynamic simulation, impeller parameter evaluation, wide-condition stability verification, and installation boundary compliance verification are re-executed on the two parameter adjustment combinations, and the new round of impeller comprehensive evaluation results are compared with the impeller comprehensive evaluation results corresponding to the current impeller reference parameter combination. When the comprehensive evaluation level of the parameter adjustment combination is higher than that of the current impeller reference parameter combination, the parameter adjustment combination is replaced with the new current impeller reference parameter combination. When the comprehensive evaluation level is the same and the weighted static pressure efficiency value is higher, the weighted flow field uniformity value is higher, the weighted separation risk value is lower, and the weighted noise proxy value is lower, the parameter adjustment combination is also replaced with the new current impeller reference parameter combination. When the parameter adjustment combination does not meet the above replacement conditions, the current impeller reference parameter combination is retained unchanged.

[0157] After completing one round of parameter perturbation and result comparison, it is further determined whether the current round's reference parameter combination has been updated. If the current round's reference parameter combination has been updated, the updated current round's reference parameter combination is used as the reference parameter combination for the next round, and parameter perturbation continues for the remaining unadjusted parameter items. If the current round's reference parameter combination has not been updated, the current perturbation step size is reduced, and parameter perturbation is re-executed for each parameter item with the reduced perturbation step size. For example, if the current round's reference parameter combination obtains a higher weighted flow field uniformity value and a lower weighted separation risk value after adjusting the blade installation angle parameter, the adjusted parameter combination is replaced with the new current round's reference parameter combination. If no better results are obtained after adjusting the splitter blade length ratio parameter and trailing edge unloading amount, the current round's reference parameter combination remains unchanged, and the parameter perturbation step size for the next round is reduced.

[0158] After completing two consecutive iterations, a convergence check is performed on the optimal impeller parameter combination for the current iteration. Specifically, the change in the co-optimization objective value corresponding to the current optimal impeller parameter combination is compared with that of the previous optimal impeller parameter combination. Then, the change in constraint pressure corresponding to the current optimal impeller parameter combination is compared with that of the previous optimal impeller parameter combination. When both the change in the co-optimization objective value and the change in constraint pressure do not exceed a preset convergence threshold, the iterative solution is terminated, and the current optimal impeller parameter combination is determined as the target impeller parameter result. When either change exceeds the preset convergence threshold, the next round of parameter perturbation and result comparison continues. For example, if the change in the co-optimization objective value corresponding to the optimal impeller parameter combination for two consecutive iterations has decreased to within the preset threshold, and the change in constraint pressure has also decreased to within the preset threshold, then the adjustment is stopped, and the current optimal impeller parameter combination is output as the target impeller parameter result.

[0159] Through the above processing, the aerodynamic thermal field noise collaborative optimization solution algorithm does not directly select a single top-ranked impeller parameter combination from the impeller comprehensive evaluation results. Instead, it first filters out feasible parameter combinations that meet the wide operating condition stability requirements and installation boundary requirements from the collaborative optimization candidate sequence. Then, it performs parameter perturbation, result comparison, and convergence judgment around each feasible parameter combination to obtain the target impeller parameter results that simultaneously meet the requirements of large air volume operation, wide operating condition stability requirements, and installation boundary constraints.

[0160] The impeller parameter evaluation refers to the process of performing aerodynamic simulation, static pressure efficiency calculation, outlet flow field uniformity calculation, low load separation risk calculation, and blade frequency noise proxy value calculation on each combination of impeller parameters in the optimizable impeller parameter set under each candidate operating condition, and performing weighted correction in combination with multi-operating condition design weights and outlet flow field constraint set to obtain the comprehensive impeller evaluation result.

[0161] The optimizable impeller parameter set refers to the set of adjustable parameters obtained after impeller design boundary transformation and impeller geometric parameter expansion based on the candidate operating condition set, benchmark impeller parameters, and installation boundary information. This set serves as the parameter input for impeller parameter evaluation and optimization. The parameters in the optimizable impeller parameter set include at least the blade installation angle parameter, blade wrap angle parameter, number of blades parameter, splitter blade length ratio parameter, blade thickness distribution parameter, leading edge sweep, and trailing edge unloading.

[0162] The parameters in the optimizable impeller parameter set are defined as follows: the blade installation angle parameter refers to the angle between the tangent of the arc in the blade profile at the inlet and the circumferential direction, denoted by the inlet installation angle. Registration, and the included angle at the exit, with the exit installation angle. Registration; the blade wrap angle parameter refers to the angle spanned by the blade inlet edge and outlet edge along the circumferential direction, in... Registration; the blade number parameter refers to the number Z of main blades evenly distributed on the impeller circumference; the splitter blade length ratio parameter refers to the ratio of the chord length of the splitter blade along the streamline direction to the chord length of the main blade along the streamline direction. The value ranges from 0.3 to 0.7; the blade thickness distribution parameter refers to the blade thickness value at each of the five control points of equal arc length along the mid-curve direction of the blade. to The thickness values ​​at each control point must meet the maximum blade thickness limit registered in the installation boundary information; the leading edge sweep refers to the maximum axial distance by which the blade's leading edge deviates from the radial reference line in the meridional projection. Offset along the mainstream direction is considered positive; trailing edge unloading refers to the blade trailing edge exit section installation angle relative to the blade mid-section reference installation angle. The amount of reduction The unit is degrees, and the value ranges from 0° to 5°.

[0163] The allowable value range of each of the above parameters is constrained by the installation boundary information registered in the design data set as the upper and lower limits. Parameter combinations that are not within the allowable range of the installation boundary will be marked as wide-condition stability failure parameter combinations and excluded in the subsequent impeller parameter evaluation.

[0164] Combination Figure 2 The diagram compares the effects of the high-flow-rate centrifugal fan impeller optimization design method with existing technologies. The black bars represent the technical effects of the present invention, while the gray bars represent existing technologies. Figure 2 It can be seen that the present invention is superior to the prior art to a certain extent.

[0165] Example 2 presents a high-flow-rate centrifugal fan impeller optimization design method based on Example 1. The steps involved in Example 1, such as design data set construction, candidate operating condition extraction, multi-operating condition design weight generation, impeller parameter construction, impeller parameter evaluation, and optimization solution processing, are still applicable in this example and will not be repeated. This example only further elaborates on the thermal field data merging, regional sensitivity level determination, and outlet partition mapping process in the outlet flow field constraint construction step. The specific scheme is as follows:

[0166] In one embodiment, the step of constructing an outlet flow field constraint set from the candidate operating condition set further includes:

[0167] First, read the candidate operating condition identifiers in the candidate operating condition set one by one, and then read the window data groups in the design data set one by one. Compare the current candidate operating condition identifier with the candidate operating condition identifiers registered in each window data group. When the two are the same, extract the terminal temperature rise deviation data, regional pressure difference fluctuation data, and reflux characterization data from the window data group. When the two are different, skip the window data group and continue reading the next window data group.

[0168] After comparing candidate operating condition identifiers, data belonging to the same candidate operating condition are sorted by region identifier, and data with the same region identifier are grouped into the same region data group. For multiple time series data in the same region data group, they are first sorted in ascending order by time window identifier; when two adjacent data have the same time window identifier and the same data item name, the data with the later time window identifier is retained, and the remaining duplicate data is deleted; when the time window identifiers are different or the data item names are different, all are retained in the sorting order.

[0169] Subsequently, the mean value of the terminal temperature rise deviation in each regional data group is calculated to obtain the representative value of the terminal temperature rise deviation for that region. The regional pressure difference value in each regional data group is then calculated by subtracting the values ​​from each adjacent time window, and the absolute values ​​of all pressure difference values ​​are then averaged to obtain the representative value of the regional pressure difference fluctuation for that region. The backflow event markers in each regional data group are counted, and the number of backflow events is divided by the total number of valid time windows corresponding to that regional data group to obtain the representative value of the backflow characterization for that region. After completing the above calculations, the candidate operating condition identifier, regional identifier, representative value of terminal temperature rise deviation, representative value of regional pressure difference fluctuation, and representative value of backflow characterization are sequentially written into the regional thermal field record. For regional data groups with a total number of valid time windows below the registration lower limit, no regional thermal field record is generated, and the regional data group is excluded from subsequent processing.

[0170] After obtaining the regional thermal field record, sensitivity correlation analysis is performed on the terminal temperature rise deviation data, the regional pressure difference fluctuation data, and the reflux characterization data to obtain a hot spot sensitivity matrix. Specifically, the hot spot temperature rise threshold, pressure difference fluctuation threshold, and reflux characterization threshold corresponding to the current unit are read from the threshold registration table, and the upper limit of the level interval corresponding to the hot spot temperature rise threshold, the pressure difference fluctuation threshold, and the reflux characterization threshold are read; then, the hot spot temperature rise threshold is subtracted from the representative value of the terminal temperature rise deviation in the current regional thermal field record to obtain the temperature rise deviation difference. When the temperature rise deviation difference is less than zero, the temperature rise sensitivity level of the region is recorded as the first level; when the temperature rise deviation difference is greater than or equal to zero, the temperature rise deviation difference is compared item by item with the upper limit of the first level interval, the upper limit of the second level interval, and the upper limit of the third level interval, and the temperature rise sensitivity level of the region is recorded as the corresponding level.

[0171] The aforementioned thresholds are derived from the thermal management design specifications of the target application area and historical statistical data from the terminal monitoring system, and are bound to the target unit identifier through the threshold registration form.

[0172] The hotspot temperature rise threshold registered in the threshold registration table is taken as 1.2 to 1.5 times the rated temperature rise of the target area, the differential pressure fluctuation threshold is taken as 5% to 10% of the rated static pressure, and the reflux characterization threshold is taken as 0.10 to 0.20. The upper limits of the first-level, second-level, and third-level intervals correspond to values ​​exceeding the threshold by 10%, 30%, and 60%, respectively. Accordingly, when the temperature rise deviation difference is less than zero, the corresponding temperature rise sensitivity level is level one; when it is less than the upper limit of the first-level interval, it corresponds to level two; when it is less than the upper limit of the second-level interval, it corresponds to level three; and when it reaches or exceeds the upper limit of the second-level interval, it corresponds to level four. The representative values ​​of differential pressure fluctuation and reflux characterization are determined according to the same classification method to correspond to the sensitivity level.

[0173] In the sensitivity registration table, temperature rise sensitivity levels 1 to 4 correspond to temperature rise sensitivity coefficients of 0.2, 0.4, 0.6, and 0.8, respectively; pressure difference sensitivity levels 1 to 4 correspond to pressure difference sensitivity coefficients of 0.15, 0.30, 0.50, and 0.70, respectively; and reflux sensitivity levels 1 to 4 correspond to reflux sensitivity coefficients of 0.10, 0.25, 0.45, and 0.65, respectively. The area sensitivity value obtained by summing the three sensitivity coefficients falls within the range of [0.45, 2.15]. When all three sensitivity levels are at the first level, the initial sensitivity value is recorded as 0.45.

[0174] Subsequently, the same interval comparison was performed on the representative values ​​of regional pressure difference fluctuations and reflux characterization, respectively, to obtain the corresponding pressure difference sensitivity level and reflux sensitivity level of the region.

[0175] After completing the three-level determination, the temperature rise sensitivity coefficient, the pressure difference sensitivity coefficient, and the reflux sensitivity coefficient corresponding to the temperature rise sensitivity level, the pressure difference sensitivity coefficient, and the reflux sensitivity coefficient corresponding to the reflux sensitivity level are read from the sensitivity registration table. The three sensitivity coefficients are then summed to obtain the area sensitivity value corresponding to the current area.

[0176] Subsequently, the candidate operating condition identifier, region identifier, and region sensitivity value are written into the hot spot sensitivity matrix. For regions where the temperature rise sensitivity level, pressure difference sensitivity level, and reflow sensitivity level are all at the first level, the corresponding region sensitivity value is recorded as the initial sensitivity value and then written into the hot spot sensitivity matrix.

[0177] After obtaining the hot spot sensitivity matrix, the hot spot sensitivity matrix is ​​subjected to outlet partition mapping and constraint quantization. Specifically: First, the total number of outlet partitions in the partition parameter registration result is read, and the wind turbine outlet annulus is continuously divided into multiple outlet partitions according to the circumferential angle, and each outlet partition is assigned a unique outlet partition identifier; the area coverage relationship table is pre-registered in the following way: first, K outlet partitions are obtained by dividing the wind turbine outlet annulus along the circumferential direction at equal angles, and each outlet partition corresponds to a central angle range of 2π / K; then, each terminal area of ​​the target unit is projected backward along the mainstream airflow direction onto the outlet annulus to obtain the projection area of ​​each terminal area on the outlet annulus; for each terminal area, the area of ​​its projection area falling into the nth outlet partition is divided by the total area of ​​the projection area to obtain the coverage ratio Cr,n of the terminal area corresponding to the nth outlet partition. For a terminal area that falls into only a single outlet partition, its coverage ratio is 1; for a terminal area that falls into multiple outlet partitions, the sum of the coverage ratios of all corresponding outlet partitions is 1.

[0178] The regional coverage relationship table records the exit partition identifier corresponding to each regional identifier and the coverage ratio of that regional identifier to each exit partition. For a region corresponding to only one exit partition, all regional sensitivity values ​​corresponding to that region are allocated to the corresponding exit partition; for a region corresponding to multiple exit partitions, the regional sensitivity values ​​corresponding to that region are allocated to each exit partition according to the coverage ratio of each exit partition.

[0179] When performing the above allocation, the formula for calculating the regional partitioning contribution value of the hot spot sensitivity inverse mapping exit non-uniform penalty constraint generation algorithm is as follows:

[0180] ,

[0181] in, This represents the regional partitioning mapping contribution value allocated from the r-th region to the n-th exit partition under the s-th candidate working condition; This represents the region sensitivity value corresponding to the r-th region under the s-th candidate working condition; s represents the coverage ratio of the r-th region corresponding to the n-th exit partition; s represents the candidate operating condition number; r represents the region number; n represents the exit partition number.

[0182] When the r-th region corresponds to only one exit partition The value is 1;

[0183] When the r-th region corresponds to multiple exit partitions, the sum of the coverage ratios of all exit partitions for that region is 1. Through the above calculation, the regional partition mapping contribution value allocated by each region to each exit partition is obtained.

[0184] After obtaining the regional partition mapping contribution value, the contribution values ​​of all regional partition mappings belonging to the same candidate working condition and corresponding to the same exit partition are accumulated to obtain the partition sensitivity record. Specifically, firstly, based on the candidate working condition identifier, all regional partition mapping contribution values ​​belonging to the same candidate working condition are grouped into the same candidate working condition group; then, based on the exit partition identifier, all regional partition mapping contribution values ​​belonging to the same candidate working condition group and corresponding to the same exit partition are grouped into the same partition contribution group; subsequently, the contribution values ​​of all regional partition mappings in the same partition contribution group are summed to obtain the partition sensitivity value corresponding to the current exit partition under the current candidate working condition. The calculation formula for the partition sensitivity value of the hot spot sensitivity reverse mapping exit non-uniform penalty constraint generation algorithm is as follows:

[0185] ,

[0186] in, This represents the partition sensitivity value corresponding to the nth exit partition under the s-th candidate operating condition; R represents the total number of regions participating in the exit mapping of the current candidate operating condition. After summing, the candidate operating condition identifier, exit partition identifier, and partition sensitivity value are written into the partition sensitivity record. For exit partitions that have not received any regional partition mapping contribution value, their partition sensitivity value is recorded as zero and then written into the partition sensitivity record.

[0187] After obtaining the partition sensitivity records, the penalty values ​​are calculated for the partition sensitivity records using the hotspot sensitivity reverse mapping exit non-uniform penalty constraint generation algorithm to obtain the non-uniform penalty constraints corresponding to each exit partition. First, the mean value of the partition sensitivity values ​​corresponding to all exit partitions under the same candidate operating condition is calculated to obtain the mean partition sensitivity under that candidate operating condition. The formula for calculating the mean partition sensitivity is:

[0188] ,

[0189] in, K represents the mean sensitivity of all exit partitions under the s-th candidate operating condition; K represents the total number of exit partitions. This indicates the exit partition number when performing a summation on all exit partitions;

[0190] After obtaining the average sensitivity of each partition, a penalty value conversion is performed on the sensitivity value corresponding to each exit partition. The penalty value calculation formula for the exit non-uniform penalty constraint generation algorithm of the hot spot sensitivity inverse mapping is as follows:

[0191] ,

[0192] in, This represents the non-uniform penalty constraint value corresponding to the nth exit partition under the sth candidate working condition; Indicates the base penalty value; Indicates the penalty coefficient for the proportion of partitions; This indicates that the sensitivity deviates from the penalty coefficient; This indicates a positive number that prevents the denominator from being zero;

[0193] in, The high-sensitivity partition determination condition means that the partition sensitivity value corresponding to the nth exit partition under the sth candidate working condition is not lower than the average partition sensitivity value corresponding to all exit partitions under the current candidate working condition; when the high-sensitivity partition determination condition is met, a sensitivity deviation penalty term is added to the non-uniform penalty constraint value corresponding to the exit partition.

[0194] in, The basic sensitive partition determination condition indicates that the partition sensitivity value corresponding to the nth exit partition under the s-th candidate working condition is lower than the average partition sensitivity value corresponding to all exit partitions under the current candidate working condition; when the basic sensitive partition determination condition is met, the non-uniform penalty constraint value corresponding to the exit partition is not increased with a sensitivity deviation penalty term.

[0195] The base penalty value The value is between 0.05 and 0.10; the partition ratio penalty coefficient. Take a value between 0.10 and 0.20; the sensitivity deviation penalty coefficient Take a value between 0.15 and 0.30; and The sum should not exceed 0.50 to avoid excessively large non-uniform penalty constraint values ​​for a single outlet zone, which could lead to a large number of impeller parameter combinations being judged as constraint mismatches.

[0196] After obtaining the non-uniform penalty constraints corresponding to each exit partition, the non-uniform penalty constraints corresponding to each exit partition are partitioned, combined, and encapsulated to complete the construction of the exit flow field constraint set containing the non-uniform penalty constraints of each exit partition. Specifically, firstly, the non-uniform penalty constraints corresponding to all exit partitions are divided into multiple candidate operating condition constraint groups according to the candidate operating condition identifier; then, according to the order of the exit partition identifiers, all non-uniform penalty constraints in the same candidate operating condition constraint group are arranged item by item.

[0197] Subsequently, the candidate operating condition identifier, outlet partition identifier, and non-uniform penalty constraint value are sequentially written into the outlet flow field constraint item, and all outlet flow field constraint items belonging to the same candidate operating condition are grouped into the same candidate operating condition constraint set. Finally, the constraint sets of all candidate operating conditions are summarized to obtain the outlet flow field constraint set containing the non-uniform penalty constraints of each outlet partition. For the outlet partition with a partition sensitivity value of zero, its non-uniform penalty constraint value is recorded as the base penalty value and then written into the outlet flow field constraint set.

[0198] In one embodiment, the export non-uniform penalty constraint generation algorithm further includes:

[0199] In an example scenario, the target unit corresponds to four terminal regions: the near-end region of the first cabinet column, the far-end region of the first cabinet column, the near-end region of the second cabinet column, and the far-end region of the second cabinet column. The fan outlet annulus is divided into eight outlet zones circumferentially. First, the terminal temperature rise deviation data, regional pressure difference fluctuation data, and recirculation characterization data corresponding to the current candidate operating condition are read from the design data set according to the candidate operating condition identifier, and then assigned to the four terminal regions according to the region identifier. Subsequently, the terminal temperature rise deviation value, regional pressure difference fluctuation value, and recirculation characterization value corresponding to the near-end region of the first cabinet column are merged to obtain the representative values ​​of terminal temperature rise deviation, regional pressure difference fluctuation, and recirculation characterization for that region. Then, the corresponding regional thermal field records are obtained for the other three terminal regions in the same way.

[0200] After obtaining the thermal field records for each region, a level determination is performed on the representative values ​​of the terminal temperature rise deviation, regional pressure difference fluctuation, and return flow characterization for each region. For example, if the representative value of the terminal temperature rise deviation for the far-end region of the first cabinet column is higher than the hot spot temperature rise threshold, and the representative value of the regional pressure difference fluctuation is higher than the pressure difference fluctuation threshold, while the representative value of the return flow characterization is higher than the return flow characterization threshold, then the temperature rise sensitivity level, pressure difference sensitivity level, and return flow sensitivity level for that region are all recorded as an upgraded level. If the representative value of the terminal temperature rise deviation for the near-end region of the second cabinet column is not higher than the hot spot temperature rise threshold, and the representative values ​​of the regional pressure difference fluctuation and return flow characterization are both within the basic range, then the three sensitivity levels for that region are all recorded as basic levels. Subsequently, the sensitivity mapping table is read, and the temperature rise sensitivity level, pressure difference sensitivity level, and return flow sensitivity level for each region are combined and mapped to regional sensitivity values, and the regional sensitivity values ​​for each region are written into the hot spot sensitivity matrix.

[0201] After obtaining the hotspot sensitivity matrix, a reverse mapping from region to exit partition is performed. For example, if the far-end region of the first rack column corresponds to the third and fourth exit partitions in the region coverage table, and the coverage ratios are 0.6 and 0.4 respectively, then the region sensitivity value corresponding to this region is allocated to the third and fourth exit partitions at 0.6 and 0.4 respectively. If the near-end region of the second rack column corresponds only to the sixth exit partition, then all the region sensitivity values ​​corresponding to this region are allocated to the sixth exit partition. If both the near-end region of the first rack column and the far-end region of the second rack column correspond to the fourth exit partition, then the sensitivity components allocated to the fourth exit partition are summed to obtain the partition sensitivity value corresponding to the fourth exit partition. After allocating and summing all regions in the above manner, the partition sensitivity record corresponding to each exit partition is obtained.

[0202] After obtaining the partition sensitivity records, a penalty level conversion is performed on the partition sensitivity values ​​corresponding to each exit partition. For example, if the partition sensitivity values ​​corresponding to the third and fourth exit partitions are higher than the average partition sensitivity of all exit partitions under the current candidate operating condition, then the third and fourth exit partitions are recorded as high-sensitivity exit partitions, and a deviation penalty value is added to the basic penalty value to obtain the corresponding higher non-uniform penalty constraint; if the partition sensitivity value corresponding to the sixth exit partition is lower than the average partition sensitivity, then the sixth exit partition is recorded as a basic-sensitivity exit partition, and only the basic penalty value and the percentage penalty value are retained. Subsequently, the non-uniform penalty constraints corresponding to each exit partition are partitioned and encapsulated according to the candidate operating condition identifier and the exit partition identifier to obtain the exit flow field constraint item corresponding to the current candidate operating condition; then, the exit flow field constraint items corresponding to all candidate operating conditions are summarized to obtain the exit flow field constraint set.

[0203] Through the above processing, the temperature rise deviation, pressure difference fluctuation, and backflow effects originally distributed in each terminal cabinet area are transformed into non-uniform penalty constraints for each outlet zone that can directly affect the impeller parameter evaluation process. Subsequently, when performing impeller parameter evaluation, if the velocity distribution of the third and fourth outlet zones corresponding to a certain impeller parameter combination deviates from the corresponding non-uniform penalty constraints in the outlet flow field constraint set, a penalty correction is performed on the impeller parameter combination. If a certain impeller parameter combination meets the corresponding constraint requirements in all highly sensitive outlet zones, the evaluation result corresponding to the impeller parameter combination is retained and entered into the subsequent optimization solution process.

[0204] The foregoing has shown and described the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The embodiments and descriptions in the specification are merely preferred examples and are not intended to limit the invention. Various changes and modifications can be made to the invention without departing from its spirit and scope, and all such changes and modifications fall within the scope of the present invention as claimed. The scope of protection of the present invention is defined by the appended claims and their equivalents.

Claims

1. A method for optimizing the design of a high-flow-rate centrifugal fan impeller, characterized in that, include: Obtain the impeller optimization design input data and perform consistency preprocessing to obtain the design data set; Candidate operating conditions are extracted from the design data set to obtain a candidate operating condition set; The impeller parameters are constructed from the candidate operating condition set to obtain an optimizable impeller parameter set; The outlet flow field constraint is constructed by performing an outlet flow field constraint on the candidate working condition set to obtain the outlet flow field constraint set; The impeller parameter evaluation is performed on the optimizable impeller parameter set and the outlet flow field constraint set to obtain the comprehensive impeller evaluation result; The comprehensive evaluation results of the impeller are optimized and solved to obtain the target impeller parameters.

2. The method for optimizing the design of a high-flow-rate centrifugal fan impeller according to claim 1, characterized in that, Candidate operating conditions are extracted from the design data set to obtain a candidate operating condition set, including: The design data set is sliced ​​according to a unified time window to obtain a set of working condition segments; Interval statistics are performed on the air volume data, static pressure data, and temperature rise data in the set of operating condition segments to obtain operating condition characteristic records; Representative working conditions are selected from the working condition feature records to obtain a candidate working condition set.

3. The method for optimizing the design of a high-flow-rate centrifugal fan impeller according to claim 2, characterized in that, The representative operating condition screening of the aforementioned operating condition characteristic records includes: The load fluctuation spectrum is obtained by performing fluctuation statistics on the load change amplitude, duration and frequency of occurrence in the aforementioned operating condition characteristic records; Clustering and merging the fluctuation intervals in the load fluctuation spectrum yields representative operating condition categories. The weights of representative working condition categories driven by load fluctuation spectrum are calculated by a multi-condition design weight generation algorithm to determine the multi-condition design weights of each candidate working condition in the candidate working condition set.

4. The method for optimizing the design of a high-flow-rate centrifugal fan impeller according to claim 1, characterized in that, The impeller parameters are constructed from the candidate operating condition set to obtain an optimizable impeller parameter set, including: Read the reference impeller parameters and installation boundary information from the design data set; Perform impeller design boundary transformation on each candidate operating condition in the candidate operating condition set to obtain impeller design boundary parameters; The impeller geometry parameters are expanded by the reference impeller parameters, the installation boundary information, and the impeller design boundary parameters to obtain a set of optimizable impeller parameters.

5. The method for optimizing the design of a high-flow-rate centrifugal fan impeller according to claim 1, characterized in that, The candidate operating condition set is used to construct the outlet flow field constraint, resulting in an outlet flow field constraint set, including: Read the terminal temperature rise deviation data, regional pressure difference fluctuation data, and reflux characterization data corresponding to the candidate operating condition set from the design data set; Sensitivity correlation analysis was performed on the terminal temperature rise deviation data, the regional pressure difference fluctuation data, and the reflux characterization data to obtain the hot spot sensitivity matrix; The hot spot sensitivity matrix is ​​subjected to outlet partition mapping and constraint quantization to obtain an outlet flow field constraint set containing non-uniform penalty constraints for each outlet partition.

6. The method for optimizing the design of a high-flow-rate centrifugal fan impeller according to claim 5, characterized in that, The hotspot sensitivity matrix is ​​subjected to exit partition mapping and constraint quantization, including: The sensitivity values ​​of each region in the hot spot sensitivity matrix are matched with the corresponding partitions of the fan outlet annular surface to obtain the partition sensitivity records. The non-uniform penalty constraint generation algorithm for hot spot sensitivity inverse mapping is used to calculate the penalty value of the partition sensitivity record to obtain the non-uniform penalty constraint corresponding to each exit partition. The non-uniform penalty constraints corresponding to each exit partition are combined and encapsulated to complete the construction of the exit flow field constraint set containing the non-uniform penalty constraints of each exit partition.

7. The method for optimizing the design of a high-flow-rate centrifugal fan impeller according to claim 1, characterized in that, The impeller parameter evaluation is performed on the optimizable impeller parameter set and the outlet flow field constraint set to obtain the comprehensive impeller evaluation result, including: Aerodynamic simulations were performed on each impeller parameter combination in the optimizable impeller parameter set under each candidate working condition to obtain the working condition aerodynamic performance record. The static pressure efficiency, outlet flow field uniformity, low load separation risk, and blade frequency noise proxy value are calculated from the aerodynamic performance records of the aforementioned operating conditions to obtain the sub-operating condition evaluation records. The evaluation records for each working condition are weighted and summarized according to the design weights of the multi-working-condition system, and then comprehensively evaluated in conjunction with the non-uniform penalty constraints of each outlet zone in the outlet flow field constraint set to obtain the comprehensive evaluation result of the impeller.

8. The method for optimizing the design of a high-flow-rate centrifugal fan impeller according to claim 7, characterized in that, The evaluation records for each working condition are weighted and summarized according to the multi-working-condition design weights, and a comprehensive evaluation is performed in conjunction with the non-uniform penalty constraints of each outlet zone in the outlet flow field constraint set, including: The evaluation records for each working condition are weighted and summarized according to the multi-condition design weights corresponding to each candidate working condition to obtain a weighted evaluation record. The weighted evaluation record is penalized and corrected according to the non-uniform penalty constraint of each outlet partition in the outlet flow field constraint set to obtain the corrected evaluation record. The revised evaluation records are comprehensively sorted to determine the overall evaluation result of the impeller.

9. The method for optimizing the design of a high-flow-rate centrifugal fan impeller according to claim 1, characterized in that, The comprehensive evaluation results of the impeller are subjected to optimization and solution processing to obtain the target impeller parameter results, including: The candidate parameter sequence is obtained by prioritizing the various impeller parameter combinations in the comprehensive evaluation results of the impeller. The candidate parameter sequence is subjected to installation boundary compliance verification and outlet flow field constraint compliance verification to obtain a feasible parameter sequence; The feasible parameter sequence is converged and iterated to obtain the target impeller parameter results.

10. The method for optimizing the design of a high-flow-rate centrifugal fan impeller according to claim 9, characterized in that, Prioritize the various impeller parameter combinations in the comprehensive impeller evaluation results, including: The various impeller parameter combinations in the comprehensive evaluation results of the impeller are ranked collaboratively according to the static pressure efficiency index, the outlet flow field uniformity index, the low load separation risk index, and the blade frequency noise proxy index to obtain a collaborative optimization candidate sequence; The collaborative optimization candidate sequence is subjected to wide-condition stability verification and installation boundary compliance verification to obtain a feasible parameter sequence; The feasible parameter sequence is iteratively solved using an aerodynamic thermal field noise collaborative optimization solution algorithm to obtain the target impeller parameter results.

Images