Fan system performance prediction method and device

Abstract

The invention belongs to the field of fan system performance detection, and discloses a fan system performance prediction method and device, and the method comprises the following steps: (1) carrying out the orthogonal design of the geometric parameters and rotating speed of a fan, obtaining the performance parameters through simulation calculation or experimental testing, and building a training set; then, on the basis of training set data, an approximate model of the relation between performance parameters and geometric parameters and the relation between the performance parameters and the rotating speed is constructed through kriging or an artificial neural network; and (2) for a fan system to be subjected to performance prediction, measuring the geometric parameters and the rotating speed of the fan system, and inputting the geometric parameters and the rotating speed into the approximate model to obtain a performance parameter prediction result. According to the method, the approximate model of the relation between the performance parameters and the geometric parameters and the rotating speed is constructed through kriging or an artificial neural network, and the fan performance can be predicted; and moreover, field online correction can be carried out on the actually-manufactured fan system, and the fan system with the air volume and the noise meeting the targets is obtained.

Description

Technical Field

[0001] The present invention belongs to the field of fan system performance detection, and more specifically, relates to a fan system performance prediction method and device. Background Art

[0002] Due to the influence of many factors such as the production, processing and transportation of the fan system, the actual geometric parameters are quite different from the design parameters, resulting in the failure of the fan system to meet the design requirements for noise and aerodynamic performance when it is installed in the overall product and equipment as a component. Currently, experience judgment or sampling is mainly used to install it in the whole machine product, and a large number of experiments are conducted to verify the influence of specific parameter changes on aerodynamic performance and noise performance. However, there are the following shortcomings:

[0003] 1. Because there are many parameters, it is impossible to determine the key geometric parameters that have a greater impact on performance.

[0004] 2. The R&D efficiency is low, and a large number of parts will be scrapped during the process.

[0005] 3. For performance deviations that are small and cannot be corrected through control algorithms, etc., the performance requirements of the entire product must be ensured. Summary of the invention

[0006] In view of the above defects or improvement needs of the prior art, the purpose of the present invention is to provide a method and device for predicting the performance of a fan system, which constructs an approximate model of the relationship between performance parameters and geometric parameters and rotational speed through kriging or artificial neural networks. The fan performance (including air volume and noise) can be predicted by measuring the geometric parameters and rotational speed of the fan system; and, using this model, the fan system actually prepared can be corrected on-site online to obtain a fan system with air volume and noise that meet the targets.

[0007] To achieve the above object, according to one aspect of the present invention, a method for predicting performance of a fan system is provided, characterized in that it comprises the following steps:

[0008] (1) Orthogonal design is performed on the geometric parameters and speed of the fan, and performance parameters are obtained through simulation calculation or experimental test to establish a training set; any set of data in the training set includes the geometric parameters, speed and corresponding performance parameters of the fan system; then, based on the training set data, an approximate model of the relationship between the performance parameters and the geometric parameters and speed is constructed through kriging or artificial neural network;

[0009] The performance parameter includes at least one of air volume and noise;

[0010] (2) For the fan system whose performance is to be predicted, its geometric parameters and rotation speed are measured, and the approximate model obtained in step (1) is input to obtain the performance parameter prediction results.

[0011] As a further preferred embodiment of the present invention, in step (1), the data of the training set is obtained under standard operating conditions; the performance parameter prediction results obtained in step (2) also correspond to standard operating conditions;

[0012] Step (2) also includes detecting the environmental parameters of the fan system for which performance prediction is to be performed, converting the environmental parameters to standard operating conditions corresponding to the environmental parameters, and then predicting the performance parameters under the standard operating conditions; wherein the environmental parameters also include temperature, air pressure, and altitude.

[0013] As a further preferred embodiment of the present invention, the approximate model of the relationship between the performance parameter, the geometric parameter and the rotation speed has also been verified for reliability, and the accuracy meets the preset requirements;

[0014] Preferably, the preset requirement is that the accuracy is greater than 95%.

[0015] As a further preferred embodiment of the present invention, the geometric parameters also include the volute profile, blade inlet angle, blade outlet angle, blade arc radius, impeller height, gap between impeller and volute, width of volute outlet, and length of volute outlet.

[0016] According to another aspect of the present invention, the present invention provides a wind turbine system performance prediction device, characterized in that it includes:

[0017] Image recognition module, used to collect geometric parameters of the fan system;

[0018] Speed ​​collection module, used to collect the speed of the fan system;

[0019] An approximate model of the relationship between performance parameters, geometric parameters and rotational speed is used to obtain and output the prediction results of fan performance parameters based on the collected fan geometric parameters and fan rotational speed; the performance parameters include at least one of air volume and noise; the approximate model of the relationship between the performance parameters, geometric parameters and rotational speed is constructed based on training set data through kriging or artificial neural network; the training set is orthogonally designed for the geometric parameters and rotational speed of the fan, and the performance parameters are established through simulation calculation or experimental test; any set of data in the training set simultaneously includes the geometric parameters, rotational speed and corresponding performance parameters of the fan system.

[0020] As a further preferred embodiment of the present invention, there are several image recognition modules, which are respectively arranged on the front, side and outlet of the fan.

[0021] According to another aspect of the present invention, the present invention provides an on-site online correction method for a fan system, characterized in that it comprises the following steps:

[0022] S1. Orthogonal design is performed on the geometric parameters and speed of the fan, and performance parameters are obtained through simulation calculation or experimental test to establish a training set; any set of data in the training set includes the geometric parameters, speed and corresponding performance parameters of the fan system; then, based on the training set data, an approximate model of the relationship between the performance parameters and the geometric parameters and speed is constructed through kriging or artificial neural network;

[0023] The performance parameters also include air volume and noise;

[0024] S2. Prepare a fan system according to the fan system design geometric parameters preset by the parameter values, measure the actual geometric parameters, and then compare the measured geometric parameters with the design geometric parameters. If the deviation between the two exceeds the preset deviation requirement, correct the prepared geometric parameters of the fan system and update the measured geometric parameters until the deviation between the measured geometric parameters and the design geometric parameters meets the preset deviation requirement;

[0025] S3. Collect the speed of the fan system, and predict the air volume prediction value and the noise prediction value by using the approximate model obtained in step S1 according to the measured geometric parameters and the collected speed;

[0026] If the predicted air volume value is less than the preset target air volume value, and the predicted noise value is also less than the preset target noise value, then: keep the geometric parameters unchanged, increase the speed of the fan system, and use the approximate model obtained in step S1 again to predict so that the predicted air volume value is equal to the preset target air volume value, and update the predicted air volume value and the predicted noise value; then, if the predicted noise value is less than the preset target noise value, the correction ends; if the predicted noise value is greater than the preset target noise value, the fan system is corrected by adding noise reduction materials;

[0027] If the predicted air volume value is greater than or equal to the preset target air volume value, and the predicted noise value is also greater than or equal to the preset target noise value, then: keep the geometric parameters unchanged, reduce the speed of the fan system, and use the approximate model obtained in step S1 again to predict so that the predicted air volume value is equal to the preset target air volume value, and update the predicted air volume value and the predicted noise value; then, if the predicted noise value is less than the preset target noise value, the correction ends; if the predicted noise value is greater than the preset target noise value, the fan system is corrected by adding noise reduction materials;

[0028] If the predicted air volume value is less than the preset target air volume value, and the predicted noise value is greater than the preset target noise value, then: the fan system is unqualified and cannot be corrected;

[0029] If the predicted air volume value is greater than or equal to the preset target air volume value, and the predicted noise value is less than the preset target noise value, then: keep the geometric parameters unchanged, reduce the speed of the fan system, and use the approximate model obtained in step S1 again to predict so that the predicted air volume value is equal to the preset target air volume value.

[0030] As a further preferred embodiment of the present invention, in step S2, the geometric parameters also include the volute profile, the blade inlet angle, the blade outlet angle, the blade arc radius, the impeller height, the gap between the impeller and the volute, the width of the volute outlet, and the length of the volute outlet.

[0031] As a further preferred embodiment of the present invention, step S2 further includes:

[0032] The flatness of the volute cover plate of the prepared fan system is measured, and the measured volute cover plate flatness is compared with the designed volute cover plate flatness. If the deviation between the two exceeds the preset deviation requirement, the flatness of the volute cover plate of the prepared fan system is corrected until the deviation between the two meets the preset deviation requirement.

[0033] Compared with the prior art, the above technical solutions conceived by the present invention can achieve the following

[0034] Beneficial effects:

[0035] 1. The present invention constructs an approximate model of the relationship between the fan performance parameters, geometric parameters and rotation speed, and can predict the fan performance parameters (including air volume and noise) by collecting the fan's geometric parameters and rotation speed.

[0036] 2. The on-site online correction method of the fan system in the present invention can compare the online geometric parameter measurement with the design parameters to find the dimensions with large differences, and make on-site corrections and verifications (such as the gap between the volute and the impeller, the outlet angle, etc.) to make the geometric parameters meet the deviation requirements. Furthermore, the air volume and noise performance of the fan system can be predicted based on the measured geometric parameters and the collected rotation speed.

[0037] 3. Based on the present invention, performance parameters such as air volume and noise can be calculated in real time; by performing online qualification testing of the fan and making appropriate adjustments for correction, the testing cost, R&D cycle and scrap rate can be greatly reduced.

[0038] In summary, the present invention collects geometric parameters of the fan system, temperature, air pressure, speed and other physical parameters, combines artificial intelligence methods to establish a prediction model, and continuously corrects it. At the same time, it can timely detect fans with unqualified performance and air volume with noise, air volume, etc. as targets, and judge key geometric parameters. BRIEF DESCRIPTION OF THE DRAWINGS

[0039] Figure 1 It is a schematic diagram of the fan system. In the figure, F(s) represents the volute profile function, R represents the arc blade radius, beta1 represents the blade inlet angle, beta2 represents the blade outlet angle, n represents the impeller rotation speed, BH represents the volute outlet size, B represents the volute outlet width, H represents the volute outlet length, Bl represents the impeller height, and Bt represents the gap between the impeller and the volute.

[0040] Figure 2 It is a flow chart of the mathematical model for predicting the core performance of a fan.

[0041] Figure 3 It is a flow chart of the online prediction and correction algorithm for the core performance of the fan.

[0042] Figure 4 It is a three-dimensional schematic diagram of the online detection device for the core performance of the fan.

[0043] Figure 5 It is a side schematic diagram of the online detection device for the core performance of the fan. DETAILED DESCRIPTION

[0044] In order to make the purpose, technical solutions and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention. In addition, the technical features involved in the various embodiments of the present invention described below can be combined with each other as long as they do not conflict with each other.

[0045] Example 1

[0046] like Figure 2 As shown in the figure, the processing flow of the fan core performance prediction mathematical model of the present invention is as follows: first, in the fan system design stage, the volute profile (i.e. Figure 1 F(s) shown), blade inlet and outlet angle (i.e. Figure 1 beta1 and beta2 shown, where beta1 represents the inlet angle of the blade and beta2 represents the outlet angle of the blade), the blade arc radius (i.e. Figure 1 R shown), impeller height (i.e. Figure 1 Bl shown) and the gap with the volute (i.e. Figure 1 Bt shown), the width of the volute outlet (i.e. Figure 1 B shown), the length of the volute outlet (i.e. Figure 1 The geometric parameter dimensions of the fan system (H) shown in the figure are orthogonally designed, and then simulation calculations or experimental tests are carried out to obtain the performance parameters of air volume Q and noise LA. An approximate model between the performance parameters, geometric parameters and rotational speed n is constructed through kriging or artificial neural networks. Then, the reliability of the approximate model is verified through 3-5 sets of geometric parameters and rotational speeds of different fan systems, and the calculation accuracy (Ra) with the largest error is taken. The accuracy Ra = (model predicted data - actual data) / actual data × 100%. If the accuracy is less than or equal to 95%, the data volume is increased, and kriging or artificial neural networks are continued to be used to continuously correct the prediction model. Finally, a reduced-order prediction model of the fan core performance, geometric parameters and rotational speed is obtained, and imported into the cloud computing platform.

[0047] The final prediction model, when used, Figure 3 As shown, the following steps can be taken: When the fan system sample is tested, the online measuring device (such as Figure 4 , Figure 5 As shown, the specific component composition of the device can be found in the following description) The speed n, blade inlet angle beta1, outlet angle beta2, blade arc radius R, volute profile F(s) and other geometric parameters are tested respectively. The collected data is uploaded to the cloud, and the final prediction model is used to perform performance prediction calculations to obtain air volume Q and noise La parameters. First, the actual processing error is considered. The actual measured geometric parameters are compared with the designed geometric parameters. If the geometric parameters have a large deviation, such as beta1, the geometric parameters are corrected online on site and input into the model for prediction again. The results are compared with the preset target air volume and noise, and further regulation is performed. Specifically:

[0048] a) If the Q predicted by the model is less than the preset Qy, and the La predicted by the model is also less than the preset Ly, the geometric parameters can be kept unchanged, and Q can be ensured to be equal to Qy by increasing the rotation speed n; then, the rotation speed n at which Q is equal to Qy is selected to predict the value of La. If the La predicted by the model is less than Ly at this time, the requirement is met. If the La predicted by the model is greater than Ly at this time, it can be corrected by adding some noise reduction materials on site (the equivalent relationship between material weight and noise reduction amount can be accumulated in advance) to ensure that La is equal to Ly (the absolute value of La-Ly is less than 0.2dB(A), which can be regarded as La equals Ly; the same below).

[0049] b) If the Q predicted by the model is greater than or equal to the preset Qy, and the La predicted by the model is also greater than or equal to the preset Ly, the geometric parameters can be kept unchanged, and the rotation speed n can be reduced to ensure that Q is equal to Qy; then, the rotation speed n at which Q is equal to Qy is selected, and the value of La is predicted again. If the La predicted by the model is less than Ly at this time, the requirement is met. If the La predicted by the model is greater than Ly at this time, some noise reduction materials are added to ensure that La is equal to Ly.

[0050] c) If the Q predicted by the model is less than Qy, and the La predicted by the model is greater than the preset Ly, the fan system is unqualified;

[0051] d) If the Q predicted by the model is greater than or equal to the preset Qy, and the La predicted by the model is less than the preset Ly, the geometric parameters can be kept unchanged and the rotation speed n can be reduced to ensure that Q is equal to Qy.

[0052] In addition, the above-mentioned fan system performance online detection method can be used in conjunction with a fan core performance online detection device, such as Figure 4 , Figure 5 As shown, the online detection device for the core performance of the fan mainly includes: a table, several image recognition modules, a support frame, a temperature, air pressure, altitude environment detection module, and a speed detection module; during the detection, the fan system is placed on the table, and the following conditions are met at the same time:

[0053] (1) Several image recognition modules are respectively arranged at the air inlet, air outlet and side of the fan; among them, the image recognition module at the air inlet focuses on reconstructing the complete volute profile size and the inlet angle of the impeller blade through images; the image recognition module at the air outlet focuses on reconstructing the outlet size BH of the volute and the outlet angle and blade arc radius of the impeller blade through images; the image recognition module at the side is mainly used to check the flatness of the cover plates on both sides of the volute to ensure that the flatness meets the design requirements. Since the internal space of the fan system to be measured is limited and the structure is complex, the present invention preferably uses image recognition modules on the front, side and outlet of the fan, so that the information of key dimensions can be obtained, ensuring the accuracy of the geometric parameter measurement. In conjunction with the above-mentioned neural network data fusion analysis method, the defective quantity is screened and judged to ensure the accuracy and stability of the online measurement equipment.

[0054] (2) The temperature, air pressure, and altitude environmental monitoring modules are arranged above the table; the measured environmental parameters such as temperature, air pressure, and altitude are used to correct the operating conditions (i.e., these environmental parameters are converted to the standard operating conditions corresponding to the environmental parameters, and then the performance parameters are predicted under the standard operating conditions to reduce errors); at the same time, considering that the vibration, noise, and temperature changes during the operation of the fan may interfere with the measurement results, the monitoring of environmental parameters such as atmospheric pressure, altitude, and temperature, which is preferably arranged here, can convert all operating conditions to standard operating conditions for comparison, thereby avoiding the impact of performance differences caused by environmental differences.

[0055] (3) The speed module is set at the front of the air inlet. The speed module can use a laser speed measuring instrument. The reason why the speed needs to be measured instead of the design speed is that due to processing errors, after the geometric parameters change, the speed will change under a fixed gear program, so the actual speed needs to be collected.

[0056] It will be easily understood by those skilled in the art that the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions and improvements made within the spirit and principles of the present invention should be included in the protection scope of the present invention.

Claims

1. A method for predicting fan system performance, characterized in that: The following steps are involved: (1) Orthogonal design is performed on the geometric parameters and speed of the fan, and performance parameters are obtained through simulation calculation or experimental test to establish a training set; any set of data in the training set includes the geometric parameters, speed and corresponding performance parameters of the fan system; then, based on the training set data, an approximate model of the relationship between the performance parameters and the geometric parameters and speed is constructed through kriging or artificial neural network; The performance parameter includes at least one of air volume and noise; (2) For the fan system whose performance is to be predicted, its geometric parameters and rotation speed are measured, and the approximate model obtained in step (1) is input to obtain the performance parameter prediction results.

2. The method for predicting fan system performance according to claim 1, characterized in that: In step (1), the data of the training set is obtained under standard working conditions; the performance parameter prediction results obtained in step (2) also correspond to standard working conditions; Step (2) also includes detecting the environmental parameters of the fan system for which performance prediction is to be performed, converting the environmental parameters to standard operating conditions corresponding to the environmental parameters, and then predicting the performance parameters under the standard operating conditions; wherein the environmental parameters also include temperature, air pressure, and altitude.

3. The method for predicting fan system performance according to claim 1, characterized in that: The approximate model of the relationship between the performance parameters, geometric parameters and rotation speed has also been verified for reliability, and the accuracy meets the preset requirements; Preferably, the preset requirement is that the accuracy is greater than 95%.

4. The method for predicting wind turbine system performance according to claim 1, characterized in that: The geometric parameters also include the volute profile, the blade inlet angle, the blade outlet angle, the blade arc radius, the impeller height, the gap between the impeller and the volute, the width of the volute outlet, and the length of the volute outlet.

5. A fan system performance prediction device, characterized in that: include: Image recognition module, used to collect geometric parameters of the fan system; Speed ​​collection module, used to collect the speed of the fan system; An approximate model of the relationship between performance parameters, geometric parameters and rotational speed is used to obtain and output the prediction results of fan performance parameters based on the collected fan geometric parameters and fan rotational speed; the performance parameters include at least one of air volume and noise; the approximate model of the relationship between the performance parameters, geometric parameters and rotational speed is constructed based on training set data through kriging or artificial neural network; the training set is orthogonally designed for the geometric parameters and rotational speed of the fan, and the performance parameters are established through simulation calculation or experimental test; any set of data in the training set simultaneously includes the geometric parameters, rotational speed and corresponding performance parameters of the fan system.

6. The wind turbine system performance prediction device according to claim 5, characterized in that: There are several image recognition modules, which are respectively arranged on the front, side and outlet of the fan.

7. A method for on-site online correction of a fan system, characterized in that: The following steps are involved: S1. Orthogonal design is performed on the geometric parameters and speed of the fan, and performance parameters are obtained through simulation calculation or experimental test to establish a training set; any set of data in the training set includes the geometric parameters, speed and corresponding performance parameters of the fan system; then, based on the training set data, an approximate model of the relationship between the performance parameters and the geometric parameters and speed is constructed through kriging or artificial neural network; The performance parameters also include air volume and noise; S2. Prepare a fan system according to the fan system design geometric parameters preset by the parameter values, measure the actual geometric parameters, and then compare the measured geometric parameters with the design geometric parameters. If the deviation between the two exceeds the preset deviation requirement, correct the prepared geometric parameters of the fan system and update the measured geometric parameters until the deviation between the measured geometric parameters and the design geometric parameters meets the preset deviation requirement; S3. Collect the speed of the fan system, and predict the air volume prediction value and the noise prediction value by using the approximate model obtained in step S1 according to the measured geometric parameters and the collected speed; If the predicted air volume value is less than the preset target air volume value, and the predicted noise value is also less than the preset target noise value, then: keep the geometric parameters unchanged, increase the speed of the fan system, and use the approximate model obtained in step S1 to predict again, so that the predicted air volume value is equal to the preset target air volume value, and update the predicted air volume value and the predicted noise value; then, if the predicted noise value is less than the preset target noise value, the correction ends; If the noise prediction value is greater than the preset target noise value, the fan system is corrected by adding noise reduction materials; If the predicted air volume value is greater than or equal to the preset target air volume value, and the predicted noise value is also greater than or equal to the preset target noise value, then: keep the geometric parameters unchanged, reduce the speed of the fan system, and use the approximate model obtained in step S1 to predict again, so that the predicted air volume value is equal to the preset target air volume value, and update the predicted air volume value and the predicted noise value; then, if the predicted noise value is less than the preset target noise value, the correction ends; If the predicted noise value is greater than the preset target noise value, the fan system is corrected by adding noise reduction materials; If the predicted air volume value is less than the preset target air volume value, and the predicted noise value is greater than the preset target noise value, then: the fan system is unqualified and cannot be corrected; If the predicted air volume value is greater than or equal to the preset target air volume value, and the predicted noise value is less than the preset target noise value, then: keep the geometric parameters unchanged, reduce the speed of the fan system, and use the approximate model obtained in step S1 again to predict so that the predicted air volume value is equal to the preset target air volume value.

8. The on-site online correction method for a fan system according to claim 7, characterized in that: In step S2, the geometric parameters include the volute profile, the blade inlet angle, the blade outlet angle, the blade arc radius, the impeller height, the gap between the impeller and the volute, the width of the volute outlet, and the length of the volute outlet.

9. The on-site online correction method for a fan system according to claim 7, characterized in that: Step S2 also includes: The flatness of the volute cover plate of the prepared fan system is measured, and the measured volute cover plate flatness is compared with the designed volute cover plate flatness. If the deviation between the two exceeds The preset deviation requirement is used to correct the flatness of the volute cover of the prepared fan system. Until the deviation between the two meets the preset deviation requirement.

Images