A method and apparatus for noise reduction
By using a pre-trained noise recognition model to detect motor torque and speed in real time, and adjusting motor parameters to reduce noise, the noise problem of existing range hoods is solved, achieving efficient noise reduction without structural changes.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- VATTI CORP LTD
- Filing Date
- 2022-12-31
- Publication Date
- 2026-06-23
AI Technical Summary
Existing range hoods generate noise when the motor speed matches the vibration frequency. Current noise reduction methods require structural changes and are cumbersome to operate, making real-time adjustments difficult.
The motor's torque and speed are detected in real time by a pre-trained noise recognition model, and the noise recognition results are output. The torque and speed are adjusted according to the results to reduce noise. Noise reduction is achieved by using noise decision surface and step function algorithms.
Noise can be detected and adjusted in real time without changing the structure of the range hood, reducing the complexity and cost of noise reduction and improving noise reduction efficiency.
Smart Images
Figure CN116131678B_ABST
Abstract
Description
Technical Field
[0001] This application relates to the field of variable frequency smoke machine technology, and in particular to a method and apparatus for noise reduction. Background Technology
[0002] Currently, range hoods are prone to generating noise during use. When the range hood motor rotates too fast or its vibration frequency matches the range hood's vibration frequency, the range hood will produce noise, causing significant inconvenience to users.
[0003] Existing technologies reduce noise generated by excessively high motor speeds by adding sound-absorbing cotton inside the range hood, but the sound insulation effect of the cotton is poor. Changing the internal structure of the range hood alters the vibration frequency of the motor, making it different from the range hood's own vibration frequency. However, each structural change only adjusts the motor's vibration frequency once. This method cannot adjust the motor's vibration frequency in real time for noise reduction, is difficult, cumbersome, and costly. Therefore, there is an urgent need for a noise reduction method that does not require structural modifications. Summary of the Invention
[0004] Therefore, it is necessary to provide a noise reduction method and apparatus to address the aforementioned technical problems.
[0005] Firstly, a noise reduction method is provided, the method comprising:
[0006] Obtain the current torque and current speed of the motor in the range hood;
[0007] The current torque and the current speed are input into a pre-trained noise recognition model, and the first noise recognition result is output.
[0008] If the first noise identification result is 0, a noise warning message is output to prompt the user to adjust the torque and speed of the motor; the first noise identification result being 0 indicates that the noise generated by the motor based on the current torque and the current speed is greater than a preset noise threshold.
[0009] As an optional implementation, the method further includes:
[0010] If the first noise identification result is 1, then the motor is controlled to work according to the current torque and the current speed; if the first noise identification result is 1, it means that the noise generated by the motor based on the current torque and the current speed is less than or equal to the preset noise threshold.
[0011] As an optional implementation, the method further includes:
[0012] Obtain a training sample set, which includes multiple training samples and the sample noise recognition results corresponding to each training sample;
[0013] Based on each training sample and the corresponding sample noise recognition result, the initial noise recognition model is trained to obtain the trained noise recognition model.
[0014] As an optional implementation, the step of training an initial noise recognition model based on each of the training samples and the corresponding sample noise recognition results to obtain a trained noise recognition model includes:
[0015] Each of the training samples is input into the initial noise recognition model, and the second noise recognition result corresponding to each of the training samples is output.
[0016] The average of the absolute values of the differences between each of the second noise recognition results and the corresponding sample noise recognition results is determined as the first recognition accuracy of the initial noise recognition model;
[0017] If the first recognition accuracy is less than the preset training accuracy threshold, then the initial noise recognition model is determined to have completed training.
[0018] If the first recognition accuracy is greater than or equal to the preset training accuracy threshold, then according to the preset model parameter adjustment strategy, the model parameters of the initial noise recognition model are adjusted, and the steps of inputting each training sample into the initial noise recognition model and outputting the second noise recognition result corresponding to each training sample are repeated until the first recognition accuracy of the initial noise recognition model is less than the preset training accuracy threshold.
[0019] As an optional implementation, adjusting the model parameters of the initial noise recognition model according to a preset model parameter adjustment strategy includes:
[0020] Obtain the preset model parameter adjustment step size, the current training round corresponding to the initial noise recognition model, and the current number of training iterations in the current training round;
[0021] In the pre-stored correspondence between the number of training iterations, the number of adjustment layers, and the adjustment coefficient, query the target number of adjustment layers and the target adjustment coefficient corresponding to the current number of training iterations;
[0022] The model parameter correction parameter is determined based on the expression as follows:
[0023] C0=K*n*Δd
[0024] Where C0 represents the model parameter correction parameter, K represents the target adjustment coefficient, n represents the current training round, and Δd represents the model parameter adjustment step size.
[0025] Based on the model parameter correction parameters, the model parameters in the target adjustment layer are adjusted.
[0026] As an optional implementation, the method further includes:
[0027] Create a decision surface regarding the torque and speed of the motor;
[0028] The torque and speed of the motor are input into the pre-trained noise recognition model. After passing through the internal algorithm and step function algorithm in the noise recognition model, the final decision value is output.
[0029] If the final decision value is 0, a noise warning message is output; the final decision value of 0 indicates that the point determined by the two values of motor torque and motor speed falls outside the decision surface area, and the final decision value of 1 indicates that the point determined by the two values of motor torque and motor speed falls within the decision surface area.
[0030] As an optional implementation, the method further includes:
[0031] Obtain a detection sample set, wherein the training sample set includes multiple detection samples and the detection noise recognition results corresponding to each detection sample;
[0032] Each of the detected samples is input into the trained initial noise recognition model, and the third noise recognition result corresponding to each of the detected samples is output.
[0033] The average of the absolute values of the differences between the third noise recognition result and the detection noise recognition result corresponding to each of the detection samples is determined as the second recognition accuracy of the initial noise recognition model;
[0034] If the second recognition accuracy is less than the preset training accuracy threshold, then the trained initial noise recognition model is determined to meet the preset training requirements.
[0035] If the second recognition accuracy is greater than or equal to the preset training accuracy threshold, then the initial noise recognition model after training is determined to not meet the preset training requirements.
[0036] Secondly, a noise reduction device is provided, the device comprising:
[0037] The first acquisition module is used to acquire the current torque and current speed of the motor in the range hood;
[0038] The input module is used to input the current torque and the current speed into a pre-trained noise recognition model and output a first noise recognition result;
[0039] The adjustment module is used to output a noise prompt message if the first noise identification result is 0, so that the user can adjust the torque and speed of the motor; the first noise identification result of 0 indicates that the noise generated by the motor based on the current torque and the current speed is greater than a preset noise threshold.
[0040] As an optional implementation, the device further includes:
[0041] The control module is configured to control the motor to operate according to the current torque and the current speed if the first noise identification result is 1; if the first noise identification result is 1, it indicates that the noise generated by the motor based on the current torque and the current speed is less than or equal to the preset noise threshold.
[0042] As an optional implementation, the device further includes:
[0043] The second acquisition module is used to acquire a training sample set, which includes multiple training samples and sample noise recognition results corresponding to each training sample.
[0044] The training module is used to train the initial noise recognition model based on each training sample and the corresponding sample noise recognition result to obtain the trained noise recognition model.
[0045] As an optional implementation, the training module is specifically used for:
[0046] Each of the training samples is input into the initial noise recognition model, and the second noise recognition result corresponding to each of the training samples is output.
[0047] The average of the absolute values of the differences between each of the second noise recognition results and the corresponding sample noise recognition results is determined as the first recognition accuracy of the initial noise recognition model;
[0048] If the first recognition accuracy is less than the preset training accuracy threshold, then the initial noise recognition model is determined to have completed training.
[0049] If the first recognition accuracy is greater than or equal to the preset training accuracy threshold, then according to the preset model parameter adjustment strategy, the model parameters of the initial noise recognition model are adjusted, and the steps of inputting each training sample into the initial noise recognition model and outputting the second noise recognition result corresponding to each training sample are repeated until the first recognition accuracy of the initial noise recognition model is less than the preset training accuracy threshold.
[0050] As an optional implementation, the training module is specifically used for:
[0051] Obtain the preset model parameter adjustment step size, the current training round corresponding to the initial noise recognition model, and the current number of training iterations in the current training round;
[0052] In the pre-stored correspondence between the number of training iterations, the number of adjustment layers, and the adjustment coefficient, query the target number of adjustment layers and the target adjustment coefficient corresponding to the current number of training iterations;
[0053] Based on the target adjustment coefficient, the current training round, and the model parameter adjustment step size, determine the model parameter correction parameters;
[0054] Based on the model parameter correction parameters, the model parameters in the target adjustment layer are adjusted.
[0055] As an optional implementation, the device further includes:
[0056] A module is created to generate a decision surface regarding the torque and speed of the motor;
[0057] The input module is used to input the torque and speed of the motor into the pre-trained noise recognition model, and output the final decision value after passing through the internal algorithm and step function algorithm in the noise recognition model;
[0058] The output module is used to output a noise warning message if the final decision value is 0; the final decision value of 0 indicates that the point determined by the two values of motor torque and motor speed falls outside the decision surface area, and the final decision value of 1 indicates that the point determined by the two values of motor torque and motor speed falls within the decision surface area.
[0059] As an optional implementation, the device further includes:
[0060] The third acquisition module is used to acquire a detection sample set, wherein the training sample set includes multiple detection samples and the detection noise recognition result corresponding to each detection sample;
[0061] The output module is used to input each of the detection samples into the trained initial noise recognition model and output the third noise recognition result corresponding to each of the detection samples.
[0062] The determining module is used to determine the average of the absolute values of the differences between the third noise recognition result and the detection noise recognition result corresponding to each of the detection samples as the second recognition accuracy of the initial noise recognition model;
[0063] The first determination module is used to determine that the trained initial noise recognition model meets the preset training requirements if the second recognition accuracy is less than the preset training accuracy threshold.
[0064] The second determination module is used to determine that the trained initial noise recognition model does not meet the preset training requirements if the second recognition accuracy is greater than or equal to the preset training accuracy threshold.
[0065] This application provides a noise reduction method and apparatus. The technical solution provided by the embodiments of this application brings at least the following beneficial effects: A pre-trained noise recognition model is used to input the current torque and current speed of the motor into the pre-trained noise recognition model, and a first noise recognition result is output. Based on the first noise recognition result, it is determined whether the noise is within an acceptable range under the current torque and current speed conditions. If it is not within an acceptable range, the current torque and speed are adjusted, thereby achieving the purpose of noise reduction. The above process does not require changing the internal structure of the range hood, simplifying the complexity of noise reduction and reducing the cost of noise reduction. It can also detect and reduce noise in real time, improving the efficiency of noise reduction.
[0066] It should be understood that the above general description and the following detailed description are exemplary and explanatory only, and do not limit this application. Attached Figure Description
[0067] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.
[0068] Figure 1 This is a schematic diagram of the structure of a range hood provided in an embodiment of this application;
[0069] Figure 2 A flowchart illustrating a noise reduction method provided in an embodiment of this application;
[0070] Figure 3A schematic diagram of a noise decision surface provided in an embodiment of this application;
[0071] Figure 4 A schematic diagram of a step function provided in an embodiment of this application;
[0072] Figure 5 A schematic diagram illustrating the data processing of a noise recognition model provided in an embodiment of this application;
[0073] Figure 6 A schematic diagram of a noise reduction device provided in an embodiment of this application;
[0074] Figure 7 This is a schematic diagram of the structure of a computer device provided in an embodiment of this application. Detailed Implementation
[0075] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.
[0076] This application provides a noise reduction method that can be applied to range hoods. For example... Figure 1 As shown, the range hood includes a controller 101 and a motor 102.
[0077] The controller 101 is connected to the motor 102 and is used to obtain the current torque and current speed of the motor 102 in the range hood, and input the current torque and current speed into the pre-trained noise recognition model, and output the noise recognition result.
[0078] Motor 102 is used to generate driving torque for the range hood.
[0079] The following will describe in detail a noise reduction method provided in the embodiments of this application, with reference to specific implementation methods. Figure 2 A flowchart of a noise reduction method provided in an embodiment of this application is shown below. Figure 2 As shown, the specific steps are as follows:
[0080] Step 201: Obtain the current torque and current speed of the motor in the range hood.
[0081] In practice, the lower the motor torque and the higher the motor speed when the range hood is working, the greater the noise generated by the motor. Therefore, the noise can be reduced by adjusting the motor torque. Furthermore, when the vibration frequency of the motor matches the vibration frequency of the range hood, resonance occurs, resulting in noise. Therefore, the noise can be reduced by adjusting the motor speed to change the motor's vibration frequency. Therefore, when determining the noise level of the range hood under current operating conditions, it is necessary to first obtain the current motor torque and speed.
[0082] Step 202: Input the current torque and current speed into the pre-trained noise recognition model and output the first noise recognition result.
[0083] In implementation, technicians pre-train a noise recognition model to obtain a pre-trained model. The input parameters of the noise recognition model are the motor's torque and speed, and the output parameter is the noise recognition result. The range hood inputs the acquired current motor torque and speed into the pre-trained noise recognition model, which then outputs the first noise recognition result.
[0084] Step 203: If the first noise identification result is 0, a noise warning message is output to prompt the user to adjust the motor's torque and speed. A first noise identification result of 0 indicates that the noise generated by the motor based on the current torque and current speed is greater than a preset noise threshold.
[0085] In implementation, if the first noise identification result is 0, the range hood outputs a noise warning message to the user, prompting the user to adjust the motor's torque and speed. A first noise identification result of 0 indicates that the noise generated by the motor based on the current torque and speed exceeds a preset noise threshold. The adjustment method can be to increase the torque and speed by a preset step size, re-inputting the increased torque and speed into the noise identification model after each increase. If the first noise identification result output by the noise identification model remains 0 even with continuous increases, it indicates that increasing the speed may not result in a non-zero first noise identification result. In this case, the torque can also be decreased by a preset step size.
[0086] Furthermore, if the first noise identification result is 1, the motor is controlled to operate according to the current torque and current speed. If the first noise identification result is 1, it means that the noise generated by the motor operating based on the current torque and current speed is less than or equal to a preset noise threshold.
[0087] In implementation, if the first noise identification result is 1, the range hood control motor will operate according to the current torque and current speed. A first noise identification result of 1 indicates that the noise generated by the motor operating at the current torque and current speed is less than or equal to a preset noise threshold, and is within an acceptable range. Therefore, no adjustment of torque and speed is required.
[0088] As an optional implementation, the noise identification model is a three-layer model. Figure 3 This is a schematic diagram of a noise decision surface provided in an embodiment of this application. Figure 3 As shown, let the motor torque be A and the motor speed be B. On the two-dimensional coordinate plane, A is extended into a first triangle formed by lines 1, 2, and 3, and B is extended into a second triangle formed by lines 4, 5, and 6. Noise is actually determined by the motor torque A and the motor speed B. Therefore, on the two-dimensional coordinate plane, mapping A to x1 and B to x2 forms a decision surface. Let the region enclosed by the two triangles represent a suitable noise range, and the remaining regions represent an unsuitable noise range. Lines 1, 2, and 3 are expressed by the following expressions:
[0089] Line 1, y=ω11x1+ω12x2+b1.
[0090] Line 2, y=ω21x1+ω22x2+b2.
[0091] Line 3, y=ω31x1+ω32x2+b3.
[0092] Lines 4, 5, and 6 are represented by the following expressions:
[0093] Line 4, y=ω41x1+ω42x2+b4.
[0094] Line 5, y=ω51x1+ω52x2+b5.
[0095] Line 6, y=ω61x1+ω62x2+b6.
[0096] Wherein, ω11, ω12, ω21, ω22, ω31, ω32, ω41, ω42, ω51, ω52, ω61, and ω62 are modulation parameters, and b1, b2, b3, b4, b5, and b6 are bias parameters.
[0097] In each of the above expressions, there exists (x1, x2) such that y = 0, that is:
[0098] Line 1, ω11x1+ω12x2+b1=0.
[0099] Line 2, ω21x1+ω22x2+b2=0.
[0100] Line 3, ω31x1+ω32x2+b3=0.
[0101] Line 4, ω41x1+ω42x2+b4=0.
[0102] Line 5, ω51x1+ω52x2+b5=0.
[0103] Line 6, ω61x1+ω62x2+b6=0.
[0104] When points within the region of the first triangle formed by lines 1, 2, and 3 simultaneously satisfy:
[0105] Line 1, ω11x1+ω12x2+b1≥0.
[0106] Line 2, ω21x1+ω22x2+b2≥0.
[0107] Line 3, ω31x1+ω32x2+b3≥0.
[0108] Or when the points within the region of the second triangle formed by lines 4, 5, and 6 simultaneously satisfy:
[0109] Line 4, ω41x1+ω42x2+b4≥0.
[0110] Line 5, ω51x1+ω52x2+b5≥0.
[0111] Line 6, ω61x1+ω62x2+b6≥0.
[0112] If the decision point falls within the area of one of the two triangles, then the decision point falls outside the triangle area; otherwise, the decision point falls outside the triangle area. Simultaneously, a step function is introduced into the data processing of the noise recognition model. Figure 4 This is a schematic diagram of a step function provided in an embodiment of this application. Figure 4 As shown, when x < 0, y = 0; when x ≥ 0, y = 1. Based on the noise decision surface and the step function, the following can be derived: Figure 5 The diagram illustrates the data processing of the noise recognition model: x1 represents the motor torque, x2 represents the motor speed, and the model parameters are ω11, ω12, ω21, ω22, ω31, ω32, ω41, ω42, ω51, ω52, ω61, ω62, ω71, ω72, ω73, ω81, ω82, ω83, ω91, ω92, b1, b2, b3, b4, b5, b6, b7, b8, and b9. x1 and x2 pass through the first layer of the model, and the processing is as follows:
[0113] ω11x1+ω12x2+b1=a1, z1=φ(a1).
[0114] ω21x1+ω22x2+b2=a2, z2=φ(a2).
[0115] ω31x1+ω32x2+b3=a3, z3=φ(a3).
[0116] ω41x1+ω42x2+b4=a4, z4=φ(a4).
[0117] ω51x1+ω52x2+b5=a5, z5=φ(a5).
[0118] ω61x1+ω62x2+b6=a6, z6=φ(a6).
[0119] z1 to z6 pass through the second layer of the model, and the processing procedure is as follows:
[0120] ω71z1+ω72z2+ω73z3+b7=a7, z7=φ(a7).
[0121] ω81z4+ω82z5+ω83z6+b8=a8, z8=φ(a8).
[0122] As an optional implementation, after points (x1, x2) pass through the first layer (having already passed the step function φ(x)), if z1, z2, and z3 are all equal to 1, then points (x1, x2) fall within the triangle formed by lines 1, 2, and 3; otherwise, they fall outside the triangle. To determine if a point falls within the triangle's region, ω71z1 + ω72z2 + ω73z3 + b7 ≥ 0 must be satisfied. Let ω71 = 1, ω72 = 1, ω73 = 1, and b7 = -2.5. Then, when only z1, z2, and z3 are equal to 1, a7 ≥ 0, and after passing through the step function φ(x), z7 = 1; otherwise, when any one of z1, z2, and z3 is 0, a7 < 0, and after passing through the step function φ(x), z7 = 0.
[0123] z7 and z8 pass through the third layer of the model, and the processing procedure is as follows:
[0124] ω91z7+ω92z8+b9=a9, y=φ(a9).
[0125] As an optional implementation, after points (x1, x2) pass through layers 1 and 2 (having already passed the step function φ(x)), if either z7 or z8 equals 1, then points (x1, x2) fall within the triangle region; otherwise, they fall outside the triangle region. To determine if a point falls within the triangle, ω91z7 + ω92z8 + b9 ≥ 0 must be satisfied. Let ω91 = 1, ω92 = 1, and b9 = -0.5. Then, when either z7 or z8 equals 1, a9 ≥ 0, and after passing through the step function φ(x), y = 1; otherwise, when both z7 and z8 are 0, a9 < 0, and after passing through the step function φ(x), y = 0.
[0126] The value of y serves as the final decision value, taking the value of either 0 or 1. When y = 1, it indicates that the point determined by the motor's torque and speed falls within the triangular area, meaning it's within a suitable noise range, and no adjustment of the motor's torque is needed. When y = 0, it indicates that the point determined by the motor's torque and speed falls outside the triangular area, meaning it's within an unsuitable noise range. In this case, the motor's torque is adjusted (increased or decreased), while the motor's speed is monitored. The torque and speed are continuously fed into the noise identification model until y = 1 (i.e., within a suitable noise range).
[0127] Furthermore, the specific operational steps for training the noise recognition model are as follows.
[0128] Step 1: Obtain the training sample set, which includes multiple training samples and the corresponding sample noise recognition results for each training sample.
[0129] In implementation, a training sample set is acquired, which includes multiple training samples and the corresponding sample noise recognition results for each training sample. The training samples include torque and rotational speed.
[0130] For example, as shown in Table 1, the training sample set consists of eight training samples (torque and speed) numbered 1, 2, 3, 4, 5, 6, 7, and 8, and the corresponding sample noise identification results for each training sample. Each training sample is labeled with its serial number.
[0131]
[0132] Step 2: Based on each training sample and the corresponding sample noise recognition results, train the initial noise recognition model to obtain the trained noise recognition model.
[0133] In practice, the initial noise recognition model is trained based on each training sample and the corresponding sample noise recognition results to obtain a trained noise recognition model.
[0134] Specifically, the execution steps are as follows: based on each training sample and the corresponding sample noise recognition result, the initial noise recognition model is trained to obtain the trained noise recognition model.
[0135] Step 1: Input each training sample into the initial noise recognition model and output the second noise recognition result corresponding to each training sample.
[0136] In implementation, each training sample in the training set is input into the initial noise recognition model, and the initial noise recognition model outputs the second noise recognition result corresponding to each training sample.
[0137] For example, as shown in Table 2, Table 2 includes 8 training samples (torque and speed), the sample noise identification results and the second noise identification results corresponding to each training sample.
[0138]
[0139]
[0140] Step 2: The average of the absolute values of the differences between the second noise recognition result and the sample noise recognition result corresponding to each training sample is determined as the first recognition accuracy of the initial noise recognition model.
[0141] In practice, the absolute value of the difference between the second noise recognition result and the sample noise recognition result corresponding to each training sample is calculated, then summed, and then the average value is calculated. This average value is determined as the first recognition accuracy of the initial noise recognition model.
[0142] For example, the noise identification results of the 8 training samples are Y1, Y2, Y3, Y4, Y5, Y6, Y7, and Y8, and the second noise identification results of the 8 training samples are y1, y2, y3, y4, y5, y6, y7, and y8. The absolute values of the differences between the second noise identification results and the sample noise identification results for each training sample are: Δy1 = |y1 - Y1|; Δy2 = |y2 - Y2|; Δy3 = |y3 - Y3|; Δy4 = |y4 - Y4|; Δy5 = |y5 - Y5|; Δy6 = |y6 - Y6|; Δy7 = |y7 - Y7|; Δy8 = |y8 - Y8|. The average of the absolute values of the differences between the second noise identification results and the sample noise identification results for each training sample is: y avg =(Δy1+Δy2+Δy3+Δy4+Δy5+Δy6+Δy7+Δy8) / 8. and put y avg The first recognition accuracy rate was determined as the initial noise recognition model.
[0143] Step 3: If the first recognition accuracy is less than the preset training accuracy threshold, the initial noise recognition model is considered to have completed training.
[0144] In implementation, the initial recognition accuracy is compared with a preset training accuracy threshold. If the initial recognition accuracy is less than the preset training accuracy threshold, it indicates that the accuracy of the initial noise recognition model is already high, thus confirming that the initial noise recognition model training is complete. The preset training accuracy threshold can be 0.1. Optionally, it can be set according to the specific application requirements; this is not limited here.
[0145] Step 4: If the first recognition accuracy is greater than or equal to the preset training accuracy threshold, then adjust the model parameters of the initial noise recognition model according to the preset model parameter adjustment strategy, and repeat the steps of inputting each training sample into the initial noise recognition model and outputting the second noise recognition result corresponding to each training sample until the first recognition accuracy of the initial noise recognition model is less than the preset training accuracy threshold.
[0146] In implementation, the first recognition accuracy is compared with a preset training accuracy threshold. If the first recognition accuracy is greater than or equal to the preset training accuracy threshold, it indicates that the accuracy of the initial noise recognition model is very low, and the model parameters of the initial noise recognition model need to be adjusted. Therefore, the model parameters of the initial noise recognition model need to be adjusted according to the preset model parameter adjustment strategy. Here, the model parameters are ω (modulation parameter) and b (bias parameter). Furthermore, each training sample is input into the adjusted initial noise recognition model, and the second noise recognition result corresponding to each training sample is output, along with the recalculation of the first recognition accuracy. The first recognition accuracy is then compared with the preset training accuracy threshold. If the first recognition accuracy is still greater than the preset training accuracy threshold, the model parameters of the initial noise recognition model are further adjusted according to the preset model parameter adjustment strategy until the first recognition accuracy of the initial noise recognition model is less than the preset training accuracy threshold.
[0147] Furthermore, the specific steps for adjusting the model parameters of the initial noise recognition model according to the preset model parameter adjustment strategy are as follows.
[0148] Step 1: Obtain the preset model parameter adjustment step size, the current training round corresponding to the initial noise recognition model, and the current number of training iterations in the current training round.
[0149] In implementation, when the initial noise recognition model's first recognition accuracy is greater than or equal to the preset training accuracy threshold, it indicates that the initial noise recognition model's accuracy is low and the initial noise recognition model has not yet been fully trained. In this case, it is necessary to adjust the model parameters of the initial noise recognition model according to the preset model parameter adjustment strategy. The model parameter adjustment strategy involves adjusting the model parameters based on the model parameter adjustment step size, the current training epoch corresponding to the initial noise recognition model, and the current training iterations within the current training epoch. Each training epoch includes multiple training iterations. Optionally, each training epoch includes 14 training iterations; however, in practical applications, the number of training iterations can be set according to the actual application situation and is not limited here. Therefore, when adjusting the model parameters, it is necessary to first obtain the preset model parameter adjustment step size, the current training epoch corresponding to the initial noise recognition model, and the current training iterations within the current training epoch, so that the model parameters can be adjusted in subsequent steps.
[0150] Step 2: In the pre-stored correspondence between the number of training iterations, the number of adjustment layers, and the adjustment coefficient, query the target number of adjustment layers and the target adjustment coefficient corresponding to the current number of training iterations.
[0151] In implementation, the initial noise identification model is a multi-layer model. Technicians pre-establish the correspondence between the training iterations, the number of adjustment layers, and the adjustment coefficients of the initial noise identification model parameters. The smoke machine stores this correspondence. Based on the current training iteration, the smoke machine queries the target adjustment layer and target adjustment coefficient corresponding to the current training iteration from the aforementioned correspondence. The target adjustment coefficient can take the value of +1 or -1.
[0152] Step 3: Adjust the step size based on the target adjustment coefficient, the current training round, and the model parameters to determine the model parameter correction parameters.
[0153] In practice, the model parameter correction parameters can be determined by adjusting the target adjustment coefficient, the current training round, and the step size of the model parameters.
[0154] As an optional implementation, the formula for determining the model parameter correction parameters, based on the target adjustment coefficient, the current training epoch, and the model parameter adjustment step size, is as follows:
[0155] C0=K*n*Δd
[0156] Where C0 represents the model parameter correction parameter, K represents the target adjustment coefficient, n represents the current training round, and Δd represents the model parameter adjustment step size.
[0157] As an optional implementation, the initial noise identification model is a three-layer model, and the model parameter adjustment strategy is as follows:
[0158] ① Add n*Δd to all model parameters of the first layer, keep the model parameters of the second and third layers unchanged, and start training again.
[0159] ② Add n*Δd to all model parameters of the second layer, keep the model parameters of the first and third layers unchanged, and start training again.
[0160] ③ Add n*Δd to all model parameters of the third layer, keep the model parameters of the first and second layers unchanged, and start training again.
[0161] ④ Add n*Δd to all model parameters of the first and second layers, keep the model parameters of the third layer unchanged, and start training again.
[0162] ⑤ Add n*Δd to all model parameters of the first and third layers, keep the model parameters of the second layer unchanged, and start training again.
[0163] ⑥ Add n*Δd to all model parameters of the second and third layers, keep the model parameters of the first layer unchanged, and start training again.
[0164] ⑦ Add n*Δd to all model parameters of the first, second, and third layers, and then start training.
[0165] ⑧ Reduce all model parameters of the first layer to -n*Δd, keep the parameters of the second and third layers unchanged, and then start training again.
[0166] ⑨ Reduce all model parameters of the second layer to -n*Δd, keep the parameters of the first and third layers unchanged, and start training again.
[0167] ⑩ All model parameters in the third layer are reduced by n*Δd, while the parameters in the first and second layers remain unchanged. Training then begins.
[0168] The first and second layers have all model parameters -n*Δd, while the third layer parameters remain unchanged. Then, training begins again.
[0169] The first and third layers have all model parameters -n*Δd, the second layer parameters remain unchanged, and then training begins.
[0170] The second and third layers have all model parameters -n*Δd, while the parameters of the first layer remain unchanged, and then training begins again.
[0171] Take all the model parameters from the first, second, and third layers and subtract n*Δd, then start training again.
[0172] Among them, ①②③④⑤⑥⑦⑧⑨⑩ The current training iteration is 1, the first, second, and third layers are the model layers, +1 or -1 is the target adjustment coefficient, n is the current training iteration, and Δd is the model parameter adjustment step size.
[0173] When adjusting the model parameters of the initial noise recognition model, first obtain the preset model parameter adjustment step size, the current training epoch corresponding to the initial noise recognition model, and the current training iteration within the current training epoch. For the current training iteration, query the target number of adjustment layers and the target adjustment coefficient corresponding to the current training iteration from the pre-stored correspondence between the training iteration, the number of adjustment layers, and the adjustment coefficient. In this way, the model parameter correction parameters can be determined.
[0174] Step 4: Adjust the model parameters in the target adjustment layer based on the model parameter correction parameters.
[0175] In implementation, the sum of the initial model parameters and the model parameter correction parameters in the target adjustment layer of the initial noise identification model is determined as the model parameters in the target adjustment layer to be adjusted.
[0176] As an optional implementation, the formula for adjusting the model parameters in the target adjustment layer based on the model parameter correction parameter is as follows:
[0177] C1=C0+K*n*Δd
[0178] Where C1 represents the adjusted model parameters, C0 represents the model parameter correction parameters, K represents the target adjustment coefficient, n represents the current training round, and Δd represents the model parameter adjustment step size.
[0179] After adjusting the model parameters, the process involves inputting each training sample into the initial noise recognition model and outputting the second noise recognition result corresponding to each training sample. If the initial noise recognition model's first recognition accuracy is still greater than or equal to the preset training accuracy threshold, the current training iteration in the current training round is incremented by one, and this incremented iteration is determined as the updated current training iteration. The model parameters of the initial noise recognition model are then adjusted according to the model parameter adjustment strategy until the initial noise recognition model's first recognition accuracy is less than the preset training accuracy threshold.
[0180] Furthermore, after training the noise recognition model, it is necessary to perform testing on the trained model. The specific steps for testing are as follows.
[0181] Step 1: Obtain the detection sample set. The training sample set includes multiple detection samples and the corresponding detection noise recognition results for each detection sample.
[0182] In implementation, a detection sample set is acquired. The training sample set includes multiple detection samples and the corresponding detection noise recognition results for each detection sample. Among them, the detection samples include the motor's speed and torque.
[0183] Step 2: Input each detection sample into the trained initial noise recognition model and output the third noise recognition result corresponding to each detection sample.
[0184] In practice, each detection sample is input into the trained initial noise recognition model, and the third noise recognition result corresponding to each detection sample is output.
[0185] Step 3: The average of the absolute values of the differences between the third noise recognition result and the detection noise recognition result corresponding to each detection sample is determined as the second recognition accuracy of the initial noise recognition model.
[0186] In practice, the average of the absolute values of the differences between each third noise identification result and the corresponding detected noise identification result is determined as the second identification accuracy of the initial noise identification model.
[0187] Step 4: If the second recognition accuracy is less than the preset training accuracy threshold, then the initial noise recognition model after training is determined to meet the preset training requirements.
[0188] In practice, the second recognition accuracy is compared with the preset training accuracy threshold. If the second recognition accuracy is less than the preset training accuracy threshold, the initial noise recognition model after training is determined to meet the preset training requirements.
[0189] Step 5: If the second recognition accuracy is greater than or equal to the preset training accuracy threshold, then it is determined that the initial noise recognition model after training does not meet the preset training requirements.
[0190] In implementation, the second recognition accuracy is compared with a preset training accuracy threshold. If the second recognition accuracy is greater than or equal to the preset training accuracy threshold, the initial noise recognition model after training is determined not to meet the preset training requirements. The steps of training the initial noise recognition model based on each training sample and the corresponding sample noise recognition results are repeated to obtain a trained noise recognition model.
[0191] This application provides a noise reduction method. Using a pre-trained noise recognition model, the current torque and speed of the motor are input into the model, and a first noise recognition result is output. Based on the first noise recognition result, it is determined whether the noise level is within an acceptable range under the current torque and speed conditions. If it is not within the acceptable range, the current torque and speed are adjusted to achieve noise reduction. This process does not require changes to the internal structure of the range hood, simplifying the complexity of noise reduction and reducing costs. Furthermore, it allows for real-time noise detection and reduction, improving noise reduction efficiency.
[0192] It should be understood that, although Figure 2The steps in the flowchart are shown sequentially as indicated by the arrows, but these steps are not necessarily executed in the order indicated by the arrows. Unless otherwise specified herein, there is no strict order in which these steps are executed, and they can be performed in other orders. Figure 2 At least some of the steps in the process may include multiple steps or multiple stages. These steps or stages are not necessarily completed at the same time, but may be executed at different times. The execution order of these steps or stages is not necessarily sequential, but may be executed in turn or alternately with other steps or at least some of the steps or stages in other steps.
[0193] It is understood that the same / similar parts between the various embodiments of the methods described above in this specification can be referred to each other. Each embodiment focuses on the differences from other embodiments, and relevant parts can be referred to the description of other method embodiments.
[0194] This application also provides a noise reduction device, such as... Figure 6 As shown, the device includes:
[0195] The first acquisition module 601 is used to acquire the current torque and current speed of the motor in the range hood;
[0196] The input module 602 is used to input the current torque and the current speed into a pre-trained noise recognition model and output a first noise recognition result;
[0197] The adjustment module 603 is used to output a noise prompt message if the first noise identification result is 0, so that the user can adjust the torque and speed of the motor; the first noise identification result of 0 indicates that the noise generated by the motor based on the current torque and the current speed is greater than a preset noise threshold.
[0198] As an optional implementation, the device further includes:
[0199] The control module is configured to control the motor to operate according to the current torque and the current speed if the first noise identification result is 1; if the first noise identification result is 1, it indicates that the noise generated by the motor based on the current torque and the current speed is less than or equal to the preset noise threshold.
[0200] As an optional implementation, the device further includes:
[0201] The second acquisition module is used to acquire a training sample set, which includes multiple training samples and sample noise recognition results corresponding to each training sample.
[0202] The training module is used to train the initial noise recognition model based on each training sample and the corresponding sample noise recognition result to obtain the trained noise recognition model.
[0203] As an optional implementation, the training module 603 is specifically used for:
[0204] Each of the training samples is input into the initial noise recognition model, and the second noise recognition result corresponding to each of the training samples is output.
[0205] The average of the absolute values of the differences between the second noise recognition result and the sample noise recognition result corresponding to each training sample is determined as the first recognition accuracy of the initial noise recognition model;
[0206] If the first recognition accuracy is less than the preset training accuracy threshold, then the initial noise recognition model is determined to have completed training.
[0207] If the first recognition accuracy is greater than or equal to the preset training accuracy threshold, then according to the preset model parameter adjustment strategy, the model parameters of the initial noise recognition model are adjusted, and the steps of inputting each training sample into the initial noise recognition model and outputting the second noise recognition result corresponding to each training sample are repeated until the first recognition accuracy of the initial noise recognition model is less than the preset training accuracy threshold.
[0208] As an optional implementation, the training module 603 is specifically used for:
[0209] Obtain the preset model parameter adjustment step size, the current training round corresponding to the initial noise recognition model, and the current number of training iterations in the current training round;
[0210] In the pre-stored correspondence between the number of training iterations, the number of adjustment layers, and the adjustment coefficient, query the target number of adjustment layers and the target adjustment coefficient corresponding to the current number of training iterations;
[0211] Based on the target adjustment coefficient, the current training round, and the model parameter adjustment step size, determine the model parameter correction parameters;
[0212] Based on the model parameter correction parameters, the model parameters in the target adjustment layer are adjusted.
[0213] As an optional implementation, the device further includes:
[0214] A module is created to generate a decision surface regarding the torque and speed of the motor;
[0215] The input module is used to input the torque and speed of the motor into the pre-trained noise recognition model, and output the final decision value after passing through the internal algorithm and step function algorithm in the noise recognition model;
[0216] The output module is used to output a noise warning message if the final decision value is 0; the final decision value of 0 indicates that the point determined by the two values of motor torque and motor speed falls outside the decision surface area, and the final decision value of 1 indicates that the point determined by the two values of motor torque and motor speed falls within the decision surface area.
[0217] As an optional implementation, the device further includes:
[0218] The third acquisition module is used to acquire a detection sample set, wherein the training sample set includes multiple detection samples and the detection noise recognition result corresponding to each detection sample;
[0219] The output module is used to input each of the detection samples into the trained initial noise recognition model and output the third noise recognition result corresponding to each of the detection samples.
[0220] The determining module is used to determine the average of the absolute values of the differences between each of the third noise recognition results and the corresponding detected noise recognition results as the second recognition accuracy of the initial noise recognition model;
[0221] The first determination module is used to determine that the trained initial noise recognition model meets the preset training requirements if the second recognition accuracy is less than the preset training accuracy threshold.
[0222] The second determination module is used to determine that the trained initial noise recognition model does not meet the preset training requirements if the second recognition accuracy is greater than or equal to the preset training accuracy threshold.
[0223] This application provides a noise reduction device. Using a pre-trained noise recognition model, the current torque and speed of the motor are input into the model, and a first noise recognition result is output. Based on the first noise recognition result, it is determined whether the noise level is within an acceptable range under the current torque and speed conditions. If it is not within the acceptable range, the current torque and speed are adjusted to achieve noise reduction. This process does not require changes to the internal structure of the range hood, simplifying the complexity of noise reduction and reducing costs. Furthermore, it allows for real-time noise detection and reduction, improving noise reduction efficiency.
[0224] For specific limitations regarding the noise reduction device, please refer to the limitations on the noise reduction method above, which will not be repeated here. Each module in the aforementioned noise reduction device can be implemented entirely or partially through software, hardware, or a combination thereof. These modules can be embedded in or independent of the processor in a computer device in hardware form, or stored in the memory of a computer device in software form, so that the processor can call and execute the operations corresponding to each module.
[0225] In one embodiment, a computer device is provided, such as Figure 7 As shown, the device includes a memory and a processor. The memory stores a computer program that can run on the processor. When the processor executes the computer program, it implements the above-described image processing method steps.
[0226] In one embodiment, a computer-readable storage medium stores a computer program thereon, which, when executed by a processor, implements the steps of the image processing method described above.
[0227] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium, and when executed, it can include the processes of the embodiments of the above methods. Any references to memory, storage, databases, or other media used in the embodiments provided in this application can include non-volatile and / or volatile memory. Non-volatile memory can include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), or flash memory. Volatile memory can include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in various forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), dual data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link DRAM (SLDRAM), Rambus direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM), etc.
[0228] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.
[0229] It should also be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for display, data used for analysis, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties.
[0230] The various embodiments in this specification are described in a related manner. Similar or identical parts between embodiments can be referred to mutually. Each embodiment focuses on describing the differences from other embodiments. In particular, the system embodiments are basically similar to the method embodiments, so the description is relatively simple; relevant parts can be referred to the descriptions of the method embodiments.
[0231] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.
[0232] The embodiments described above are merely illustrative of several implementation methods of this application, and while the descriptions are relatively specific and detailed, they should not be construed as limiting the scope of the invention patent. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this application, and these all fall within the protection scope of this application. Therefore, the protection scope of this patent application should be determined by the appended claims.
Claims
1. A noise reduction method, characterized in that, The method includes: Obtain the current torque and current speed of the motor in the range hood; The current torque and the current speed are input into a pre-trained noise recognition model, and the first noise recognition result is output. If the first noise identification result is 0, a noise warning message is output to prompt the user to adjust the torque and speed of the motor; the first noise identification result being 0 indicates that the noise generated by the motor based on the current torque and current speed is greater than a preset noise threshold. The process of training the noise recognition model is as follows: A training sample set is obtained, which includes multiple training samples and the sample noise recognition results corresponding to each training sample; based on each training sample and the sample noise recognition results corresponding to each training sample, an initial noise recognition model is trained to obtain the trained noise recognition model. The process of training an initial noise recognition model based on each training sample and the corresponding sample noise recognition result to obtain a trained noise recognition model includes: Each training sample is input into the initial noise recognition model, and a second noise recognition result corresponding to each training sample is output. The average of the absolute values of the differences between each second noise recognition result and the corresponding sample noise recognition result is determined as the first recognition accuracy of the initial noise recognition model. If the first recognition accuracy is less than a preset training accuracy threshold, the initial noise recognition model is determined to be trained successfully. If the first recognition accuracy is greater than or equal to the preset training accuracy threshold, the model parameters of the initial noise recognition model are adjusted according to a preset model parameter adjustment strategy, and the steps of inputting each training sample into the initial noise recognition model and outputting the second noise recognition result corresponding to each training sample are repeated until the first recognition accuracy of the initial noise recognition model is less than the preset training accuracy threshold. Executing the preset model parameter adjustment strategy to adjust the model parameters of the initial noise recognition model includes: Obtain the preset model parameter adjustment step size, the current training epoch corresponding to the initial noise recognition model, and the current training iteration in the current training epoch; query the target adjustment epoch and target adjustment coefficient corresponding to the current training iteration from the pre-stored correspondence among the training iterations, adjustment layers, and adjustment coefficients; determine the model parameter correction parameters according to the expression: ； Where C0 represents the model parameter correction parameter, K represents the target adjustment coefficient, n represents the current training round, and Δd represents the model parameter adjustment step size; Based on the model parameter correction parameters, the model parameters in the target adjustment layer are adjusted.
2. The method according to claim 1, characterized in that, The method further includes: If the first noise identification result is 1, then the motor is controlled to work according to the current torque and the current speed; if the first noise identification result is 1, it means that the noise generated by the motor based on the current torque and the current speed is less than or equal to the preset noise threshold.
3. The method according to claim 1, characterized in that, The method further includes: Create a decision surface regarding the torque and speed of the motor; The torque and speed of the motor are input into the pre-trained noise recognition model. After passing through the internal algorithm and step function algorithm in the noise recognition model, the final decision value is output. If the final decision value is 0, a noise warning message is output; the final decision value of 0 indicates that the point determined by the two values of motor torque and motor speed falls outside the decision surface area, and the final decision value of 1 indicates that the point determined by the two values of motor torque and motor speed falls within the decision surface area.
4. The method according to claim 1, characterized in that, The method further includes: Obtain a detection sample set, which includes multiple detection samples and the detection noise identification results corresponding to each detection sample; Each of the detected samples is input into the trained initial noise recognition model, and the third noise recognition result corresponding to each of the detected samples is output. The average of the absolute values of the differences between each of the third noise recognition results and the corresponding detected noise recognition results is determined as the second recognition accuracy of the initial noise recognition model; If the second recognition accuracy is less than the preset training accuracy threshold, then the trained initial noise recognition model is determined to meet the preset training requirements. If the second recognition accuracy is greater than or equal to the preset training accuracy threshold, then the initial noise recognition model after training is determined to not meet the preset training requirements.
5. A noise reduction device, characterized in that, The device includes: The first acquisition module is used to acquire the current torque and current speed of the motor in the range hood; The input module is used to input the current torque and the current speed into a pre-trained noise recognition model and output a first noise recognition result; The adjustment module is used to output a noise warning message if the first noise identification result is 0, so that the user can adjust the torque and speed of the motor; the first noise identification result of 0 indicates that the noise generated by the motor based on the current torque and the current speed is greater than a preset noise threshold. The process of training the noise recognition model is as follows: A training sample set is obtained, which includes multiple training samples and the sample noise recognition results corresponding to each training sample; based on each training sample and the sample noise recognition results corresponding to each training sample, an initial noise recognition model is trained to obtain the trained noise recognition model. The process of training an initial noise recognition model based on each training sample and the corresponding sample noise recognition result to obtain a trained noise recognition model includes: Each training sample is input into the initial noise recognition model, and a second noise recognition result corresponding to each training sample is output. The average of the absolute values of the differences between each second noise recognition result and the corresponding sample noise recognition result is determined as the first recognition accuracy of the initial noise recognition model. If the first recognition accuracy is less than a preset training accuracy threshold, the initial noise recognition model is determined to be trained successfully. If the first recognition accuracy is greater than or equal to the preset training accuracy threshold, the model parameters of the initial noise recognition model are adjusted according to a preset model parameter adjustment strategy, and the steps of inputting each training sample into the initial noise recognition model and outputting the second noise recognition result corresponding to each training sample are repeated until the first recognition accuracy of the initial noise recognition model is less than the preset training accuracy threshold. Executing the preset model parameter adjustment strategy to adjust the model parameters of the initial noise recognition model includes: Obtain the preset model parameter adjustment step size, the current training epoch corresponding to the initial noise recognition model, and the current training iteration in the current training epoch; query the target adjustment epoch and target adjustment coefficient corresponding to the current training iteration from the pre-stored correspondence among the training iterations, adjustment layers, and adjustment coefficients; determine the model parameter correction parameters according to the expression: ； Where C0 represents the model parameter correction parameter, K represents the target adjustment coefficient, n represents the current training round, and Δd represents the model parameter adjustment step size; Based on the model parameter correction parameters, the model parameters in the target adjustment layer are adjusted.