Eccentric adjustment method of clothes treatment equipment, clothes treatment equipment and storage medium

Abstract

The embodiment of the invention discloses an eccentric adjustment method of clothes treatment equipment, the clothes treatment equipment and a storage medium, and relates to the technical field of household appliance control. The method comprises the steps that the current driving power and the current rotating speed sequence of the clothes processing equipment are collected, the current rotating speed fluctuation amplitude is determined based on the current rotating speed sequence, and the current frequency domain feature of the current rotating speed sequence is extracted; utilizing an eccentricity prediction model to predict the eccentricity degree of the clothes processing equipment at the current moment based on the current driving power, the current rotating speed fluctuation amplitude and the current frequency domain characteristics, and obtaining a target eccentricity prediction value; and determining a target balance strategy based on the target eccentricity prediction value, and executing the target balance strategy to balance the load in the clothes processing equipment, thereby improving the accuracy of eccentricity prediction, further effectively improving the eccentricity condition of the clothes processing equipment, prolonging the service life of the clothes processing equipment, protecting clothes, and improving the user experience. Therefore, safe, stable and efficient operation of the clothes processing equipment is guaranteed.

Description

Technical Field

[0001] This application relates to the field of home appliance control technology, and in particular to an eccentricity adjustment method for a clothing processing device, a clothing processing device, and a storage medium. Background Technology

[0002] Currently, the main methods for adjusting the eccentricity of garment processing equipment are numerical calculations and table lookups. These methods determine whether the equipment is eccentric based on drive power and rotation speed, and then adjust the equipment accordingly. However, these methods rely too heavily on pre-set mathematical models and fixed standard values, making it difficult to comprehensively and accurately consider complex factors such as the material, weight, absorbency, and loading method of the garments. This makes it difficult to effectively cope with various complex and changing realities, resulting in low accuracy in eccentricity prediction and consequently, low accuracy in eccentricity adjustment. Consequently, a series of problems arise during the operation of the garment processing equipment, such as poor dehydration, accelerated garment wear, excessive machine vibration, and even damage to components, reducing the lifespan of both the equipment and the garments. Summary of the Invention

[0003] This application provides an eccentricity adjustment method for a garment processing device, a garment processing device, and a storage medium, which realizes the eccentricity adjustment function of the garment processing device to solve the problem of low accuracy of eccentricity adjustment in the prior art.

[0004] In a first aspect, embodiments of this application provide a method for adjusting the eccentricity of a garment processing device. The method includes: acquiring the current driving power and current rotational speed sequence of the garment processing device; determining the current rotational speed fluctuation amplitude based on the current rotational speed sequence; and extracting the current frequency domain features of the current rotational speed sequence; using an eccentricity prediction model, predicting the degree of eccentricity of the garment processing device at the current moment based on the current driving power, current rotational speed fluctuation amplitude, and current frequency domain features, to obtain a target eccentricity prediction value; determining a target balancing strategy based on the target eccentricity prediction value, and executing the target balancing strategy to balance the load inside the garment processing device.

[0005] Secondly, embodiments of this application provide an eccentricity adjustment device for a garment processing equipment. The device includes: a data acquisition model, used to acquire the current driving power and current speed sequence of the garment processing equipment, determine the current speed fluctuation amplitude based on the current speed sequence, and extract the current frequency domain features of the current speed sequence; a prediction module, used to use an eccentricity prediction model to predict the degree of eccentricity of the garment processing equipment at the current moment based on the current driving power, current speed fluctuation amplitude, and current frequency domain features, to obtain a target eccentricity prediction value; and an execution module, used to determine a target balancing strategy based on the target eccentricity prediction value, and execute the target balancing strategy to balance the load inside the garment processing equipment.

[0006] Thirdly, embodiments of this application provide a garment processing device, which includes: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores a computer program executable by the at least one processor, and the computer program is executed by the at least one processor to enable the at least one processor to execute the eccentricity adjustment method of the garment processing device according to any embodiment of this application.

[0007] Fourthly, embodiments of this application provide a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the eccentric adjustment method of the clothing processing device as described in any embodiment of this application.

[0008] In this embodiment, the current driving power and current speed sequence of the clothing processing equipment can be collected, the current speed fluctuation amplitude can be determined based on the current speed sequence, and the current frequency domain features of the current speed sequence can be extracted; using an eccentricity prediction model, the degree of eccentricity of the clothing processing equipment at the current moment can be predicted based on the current driving power, the current speed fluctuation amplitude, and the current frequency domain features to obtain the target eccentricity prediction value; based on the target eccentricity prediction value, a target balancing strategy can be determined and executed to balance the load inside the clothing processing equipment, thereby realizing the eccentricity adjustment function of the clothing processing equipment. In the above technical solution, a pre-trained eccentricity prediction model is used to predict the degree of eccentricity of the clothing processing equipment at the current moment based on the current driving power, the current speed fluctuation amplitude, and the current frequency domain characteristics. This leverages the powerful nonlinear fitting capability of neural networks to deeply explore the complex intrinsic relationship between driving power, speed fluctuation amplitude, and the frequency domain characteristics of speed and eccentricity mass. This allows for a comprehensive and accurate consideration of complex factors such as clothing material, weight, absorbency, and loading method, effectively addressing various complex and changing realities, thereby improving the accuracy of the target eccentricity prediction. Furthermore, based on the target eccentricity prediction, a target balancing strategy is determined and executed, effectively improving the eccentricity of the clothing processing equipment, reducing wear on key components, extending the equipment's lifespan, reducing maintenance costs, improving clothing processing efficiency, protecting clothing from wear, ensuring the safe, stable, and efficient operation of the clothing processing equipment, and enhancing the user experience. Attached Figure Description

[0009] To more clearly illustrate the technical solutions in the embodiments of this application, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0010] Figure 1 This is a schematic flowchart of an eccentric adjustment method for a garment processing device provided in an embodiment of this application;

[0011] Figure 2 This is another schematic flowchart of the eccentric adjustment method for the clothing processing equipment provided in the embodiments of this application;

[0012] Figure 3 This is a structural example diagram of the eccentricity prediction model provided in the embodiments of this application;

[0013] Figure 4 This is a schematic diagram of the eccentric adjustment device of the clothing processing equipment provided in the embodiments of this application;

[0014] Figure 5 This is a schematic diagram of the clothing processing device provided in an embodiment of this application. Detailed Implementation

[0015] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of this application.

[0016] It should be noted that the terms "first," "second," "target," and "original," etc., used in the specification, claims, and accompanying drawings of this application are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of this application described herein can be implemented in sequences other than those illustrated or described herein. Furthermore, the terms "comprising," "having," and any variations thereof are intended to cover a non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0017] Figure 1This is a flowchart illustrating an eccentricity adjustment method for a garment processing device provided in this application embodiment. This embodiment can be applied to scenarios requiring prediction and adjustment of the eccentricity of the garment processing device. The eccentricity adjustment method for a garment processing device provided in this embodiment can be executed by an eccentricity adjustment device for the garment processing device provided in this application embodiment. This device can be implemented through software and / or hardware. In a specific embodiment, the eccentricity adjustment device for the garment processing device can be integrated into the overall controller of the garment processing device. For example, the garment processing device can be a washing machine, dryer, or washer-dryer combo, and it includes a high-speed rotating garment-carrying component. The executing entity for this method can be the overall controller of the garment processing device. See also... Figure 1 The eccentricity adjustment method of the garment processing equipment in this embodiment includes, but is not limited to, the following steps:

[0018] S110. Collect the current drive power and current speed sequence of the clothing processing equipment, determine the current speed fluctuation amplitude based on the current speed sequence, and extract the current frequency domain features of the current speed sequence.

[0019] Here, "Current Drive Power" refers to the actual operating power of the drive motor of the garment handling equipment at the current moment. "Current Speed ​​Sequence" is a set of real-time speed data continuously collected at preset intervals during the garment-carrying component (e.g., the drum) of the garment handling equipment as it rotates a preset number of revolutions starting from the current moment; the preset number of revolutions is a pre-set number, such as 8. "Current Speed ​​Fluctuation Amplitude" characterizes the maximum speed fluctuation in the current speed sequence. "Current Frequency Domain Feature" refers to the frequency domain characteristics of the current speed sequence.

[0020] Optionally, the current frequency domain features may include the current peak frequency and the current maximum amplitude of the spectrum; the current maximum amplitude of the spectrum is the maximum amplitude of the frequency domain signal obtained after transforming the current speed sequence through the time-domain to frequency-domain conversion algorithm, that is, the amplitude corresponding to the current peak frequency of the spectrum, reflecting the magnitude of the fluctuation corresponding to the peak frequency; the current peak frequency of the spectrum is the frequency value corresponding to the frequency component with the highest amplitude after transforming the current speed sequence through the time-domain to frequency-domain conversion algorithm, reflecting the core periodic law of speed fluctuation.

[0021] Specifically, the drive power of the drive motor of the garment processing equipment at the current moment can be collected to obtain the current drive power. Multiple rotational speed data points of the garment-carrying component (such as the roller) of the garment processing equipment, obtained by rotating a preset number of revolutions from the current moment at preset intervals, can be collected to construct the current rotational speed sequence. The preset interval angle is a pre-set value that can be adjusted and set according to actual conditions.

[0022] Then, the maximum and minimum speeds are obtained from the current speed sequence, and the difference between the maximum and minimum speeds is calculated to obtain the current speed fluctuation amplitude.

[0023] Then, the current rotational speed sequence is transformed using time-domain to frequency-domain conversion algorithms such as Fourier transform to obtain a frequency domain signal. This frequency domain signal includes a set of frequencies and their corresponding amplitude sets. Next, the current frequency domain features of the current rotational speed sequence are extracted from this frequency domain signal. For example, the maximum amplitude value is obtained from the amplitude set to obtain the current maximum amplitude value of the spectrum, and the frequency corresponding to the current maximum amplitude value of the spectrum is obtained from the frequency set to obtain the current main peak frequency of the spectrum.

[0024] S120. Using the eccentricity prediction model, the degree of eccentricity of the clothing processing equipment at the current moment is predicted based on the current driving power, the current speed fluctuation amplitude, and the current frequency domain characteristics, so as to obtain the target eccentricity prediction value.

[0025] The eccentricity prediction model is a pre-trained neural network model used to predict the degree of eccentricity of the clothing processing equipment based on the driving power, rotational speed fluctuation amplitude, and frequency domain characteristics of the rotational speed. The target eccentricity prediction value is the output of the eccentricity prediction model, which is the degree of eccentricity predicted by the eccentricity prediction model based on the current driving power, current rotational speed fluctuation amplitude, and current frequency domain characteristics.

[0026] Specifically, after obtaining the current driving power, the current speed fluctuation amplitude, and the current frequency domain features, a pre-trained eccentricity prediction model can be obtained. The current driving power, the current speed fluctuation amplitude, and the current frequency domain features are then input into the eccentricity prediction model. At this point, the eccentricity prediction model can use the learned model parameters to extract and analyze the current driving power, the current speed fluctuation amplitude, and the current frequency domain features, and predict the degree of eccentricity of the clothing processing equipment at the current moment, i.e., the target eccentricity prediction value.

[0027] S130. Determine the target balancing strategy based on the target eccentricity prediction value and execute the target balancing strategy to balance the load inside the garment processing equipment.

[0028] Among them, the balancing strategy is a series of operating procedures that adjust the load (such as clothes) inside the garment processing equipment to reduce the eccentricity of the garment processing equipment, such as the mechanical shaking program; the target balancing strategy is the balancing strategy corresponding to the target eccentricity prediction value.

[0029] Specifically, after obtaining the target eccentricity prediction value, a target balancing strategy can be determined based on the target eccentricity prediction value. For example, a preset eccentricity threshold can be obtained, where the preset eccentricity threshold is a pre-set acceptable eccentricity threshold value. If the target eccentricity prediction value is not greater than the preset eccentricity threshold, it indicates that the target eccentricity prediction value is within the acceptable eccentricity range, and no balancing strategy needs to be applied to the clothing processing equipment. If the target eccentricity prediction value is greater than the preset eccentricity threshold, it indicates that the target eccentricity prediction value is not within the acceptable eccentricity range, and the target balancing strategy can be determined based on the deviation between the target eccentricity prediction value and the preset eccentricity threshold.

[0030] For example, when the deviation value is not greater than a first set deviation value, determining the target balancing strategy includes controlling the drive motor to rotate forward and backward according to a first set frequency, a first set amplitude, and a first set duration, so that the load inside the garment processing equipment is redistributed under inertia; when the deviation value is greater than the first set deviation value but not greater than a second set deviation value, determining the target balancing strategy includes controlling the drive motor to rotate forward and backward according to a second set frequency, a second set amplitude, and a second set duration, so that the load inside the garment processing equipment is redistributed under inertia; when the deviation value is greater than the second set deviation value, determining the target balancing strategy includes controlling the drive motor to rotate forward and backward according to a third set frequency, a third set amplitude, and a third set duration, so that the load inside the garment processing equipment is redistributed under inertia. Wherein, the first set frequency is less than the second set frequency, the second set frequency is less than the third set frequency; the first set amplitude is less than the second set amplitude, the second set amplitude is less than the third set amplitude; the first set duration is less than the second set duration, and the second set duration is less than the third set duration.

[0031] After determining the target balancing strategy, it can be implemented to balance the load inside the garment processing equipment and reduce the eccentricity of the garment processing equipment.

[0032] The technical solution of this application embodiment can collect the current driving power and current speed sequence of the clothing processing equipment, determine the current speed fluctuation amplitude based on the current speed sequence, and extract the current frequency domain features of the current speed sequence; using an eccentricity prediction model, predict the degree of eccentricity of the clothing processing equipment at the current moment based on the current driving power, current speed fluctuation amplitude, and current frequency domain features, and obtain a target eccentricity prediction value; determine a target balancing strategy based on the target eccentricity prediction value, and execute the target balancing strategy to balance the load inside the clothing processing equipment, thereby realizing the eccentricity adjustment function of the clothing processing equipment. In the above technical solution, a pre-trained eccentricity prediction model is used to predict the degree of eccentricity of the clothing processing equipment at the current moment based on the current driving power, the current speed fluctuation amplitude, and the current frequency domain characteristics. This leverages the powerful nonlinear fitting capability of neural networks to deeply explore the complex intrinsic relationship between driving power, speed fluctuation amplitude, and the frequency domain characteristics of speed and eccentricity mass. This allows for a comprehensive and accurate consideration of complex factors such as clothing material, weight, absorbency, and loading method, effectively addressing various complex and changing realities, thereby improving the accuracy of the target eccentricity prediction. Furthermore, based on the target eccentricity prediction, a target balancing strategy is determined and executed, effectively improving the eccentricity of the clothing processing equipment, reducing wear on key components, extending the equipment's lifespan, reducing maintenance costs, improving clothing processing efficiency, protecting clothing from wear, ensuring the safe, stable, and efficient operation of the clothing processing equipment, and enhancing the user experience.

[0033] The following further describes an eccentricity adjustment method for a garment processing device provided in an embodiment of this application. Figure 2 This is another schematic flowchart illustrating the eccentricity adjustment method for the garment processing equipment provided in this application. This application's embodiments are optimizations based on the above embodiments. See also... Figure 2 The method in this embodiment includes, but is not limited to, the following steps:

[0034] S210. Collect the current drive power and current speed sequence of the clothing processing equipment, determine the current speed fluctuation amplitude based on the current speed sequence, and extract the current frequency domain features of the current speed sequence.

[0035] Optionally, the input features of the eccentricity prediction model are obtained by feature selection from multiple candidate features. In one implementation, the feature selection steps are as follows, including Sa1-Sa4:

[0036] Sa1. Determine the correlation between each candidate feature and the eccentricity quality to obtain the correlation of the corresponding candidate feature, and determine the importance between each candidate feature and the eccentricity quality to obtain the importance of the corresponding candidate feature.

[0037] Specifically, the correlation between candidate features and eccentricity quality can be determined using Pearson correlation analysis or Spearman correlation analysis, and the importance between candidate features and eccentricity quality can be determined using gradient boosting tree feature importance analysis. The specific implementation details can be found in the existing technologies and will not be elaborated here.

[0038] Sa2. Sort multiple candidate features in descending order of relevance to obtain a first sorting result, and determine the relevant feature set based on the first sorting result and the first preset number of candidate features.

[0039] The first ranking result is obtained by sorting multiple candidate features in descending order of relevance. The preset number is a pre-set number of features to be selected, which can be adjusted and set according to actual needs. The relevant feature set is the set of candidate features that rank within the top preset number in the first ranking result.

[0040] Sa3. Sort multiple candidate features in descending order of importance to obtain a second sorting result, and determine the important feature set based on the first preset number of candidate features in the second sorting result.

[0041] The second ranking result is obtained by sorting multiple candidate features in descending order of importance. The important feature set is the set of the top preset number of candidate features in the second ranking result.

[0042] Sa4. The intersection of the relevant feature set and the important feature set is determined as the input feature.

[0043] In this embodiment, correlation and importance analysis can improve computational efficiency and reduce implementation complexity, thereby improving the accuracy and efficiency of input feature selection. This further enables the eccentricity prediction model to better capture the complex intrinsic relationship between input features and eccentricity quality, thus improving the prediction accuracy of the eccentricity prediction model.

[0044] Optionally, the data type of the target model parameters for the eccentric prediction model is integer, such as int8. The training process for the eccentric prediction model is as follows, including Sb1-Sb5:

[0045] Sb1, Obtain the training dataset.

[0046] The training dataset may include multiple initial sample data and the corresponding eccentricity value label for each initial sample data; the eccentricity value label is the correct eccentricity value of the initial sample data, which can be collected by placing a mass block inside the garment processing equipment; the initial sample data is the data of the garment processing equipment at historical moments.

[0047] The initial sample data may include the initial sample drive power, the initial sample rotation speed fluctuation amplitude, and the initial sample frequency domain characteristics. The initial sample drive power is the drive power of the clothing processing equipment at a historical moment. The initial sample rotation speed fluctuation amplitude is the rotation speed fluctuation amplitude of the rotation speed sequence obtained from the clothing processing equipment at a historical moment. The initial sample frequency domain characteristics are the frequency domain characteristics of the rotation speed sequence obtained from the clothing processing equipment at a historical moment, including the main peak frequency of the initial sample spectrum and the maximum amplitude of the initial sample spectrum. Furthermore, the initial sample drive power, initial sample rotation speed fluctuation amplitude, and initial sample frequency domain characteristics in the same initial sample data are located at the same historical moment.

[0048] Sb2. Scale the driving power of each initial sample to obtain the driving power of the corresponding target sample, scale the speed fluctuation amplitude of each initial sample to obtain the speed fluctuation amplitude of the corresponding target sample, and scale the frequency domain features of each initial sample to obtain the frequency domain features of the corresponding target sample.

[0049] Among them, the target sample driving power is the driving power obtained after scaling the initial sample driving power, and the value range is [-1,1]; the target sample speed fluctuation amplitude is the speed fluctuation amplitude obtained after scaling the initial sample speed fluctuation amplitude, and the value range is [-1,1]; the target sample frequency domain feature is the frequency domain feature obtained after scaling the initial sample frequency domain feature, including the main peak frequency of the target sample spectrum and the maximum amplitude of the target sample spectrum, and the value range is [-1,1].

[0050] Specifically, the driving power of each initial sample can be scaled to obtain the corresponding target sample driving power. More specifically, the power mean and power standard deviation of the driving power of all initial samples in the training dataset can be calculated, where the power mean is the average of the driving power of all initial samples in the training dataset, and the power standard deviation is the standard deviation of the driving power of all initial samples in the training dataset. When the power standard deviation is a power of a preset value (i.e., 2), the driving power of each initial sample is scaled based on the power mean and power standard deviation to obtain the corresponding target sample driving power. That is, any initial sample driving power is selected as the current initial sample driving power, and the current initial sample driving power is calculated. The difference between the initial sample driving power and the mean power is calculated, and the ratio of this difference to the power standard deviation is used to obtain the target sample driving power corresponding to the current initial sample driving power. When the power standard deviation is not a power of a preset value, a power scaling factor is determined based on the power standard deviation. This power scaling factor is closest to the power standard deviation, greater than the power standard deviation, and a power of a preset value. Each initial sample driving power is then scaled based on the mean power and the power scaling factor to obtain the corresponding target sample driving power. Specifically, any initial sample driving power is selected as the current initial sample driving power, the difference between the current initial sample driving power and the mean power is calculated, and the ratio of this difference to the power scaling factor is calculated to obtain the target sample driving power corresponding to the current initial sample driving power. Through scaling, the initial sample driving power can be mapped to a target sample driving power within the range [-1, 1]. This improves computational efficiency, reduces implementation complexity, and thus improves the efficiency and accuracy of determining the target sample driving power, providing an accurate data foundation for subsequent quantization of model parameters.

[0051] Similarly, using the same scaling process as the initial sample driving power, the speed fluctuation amplitude of each initial sample is scaled to obtain the corresponding target sample speed fluctuation amplitude. Specifically, the average and standard deviation of the speed fluctuation amplitudes of all initial samples in the training dataset are calculated, where the average fluctuation amplitude is the average of the speed fluctuation amplitudes of all initial samples in the training dataset, and the standard deviation is the standard deviation of the speed fluctuation amplitudes of all initial samples in the training dataset. When the standard deviation of the fluctuation amplitude is a power of a preset value, the speed fluctuation amplitude of each initial sample is scaled based on the average and standard deviation of the fluctuation amplitude to obtain the corresponding target sample speed fluctuation amplitude. When the standard deviation of the fluctuation amplitude is not a power of a preset value, a scaling factor is determined based on the standard deviation of the fluctuation amplitude, and the speed fluctuation amplitude of each initial sample is scaled based on the average and scaling factor of the fluctuation amplitude to obtain the corresponding target sample speed fluctuation amplitude. For specific implementation details, please refer to the scaling process details for the initial sample driving power described above.

[0052] Similarly, using the same scaling process as the initial sample driving power, the frequency domain features of each initial sample are scaled to obtain the corresponding target sample frequency domain features. That is, the average frequency and standard deviation of the main peak frequency of the spectrum of all initial samples in the training dataset are calculated, where the average frequency is the average of the main peak frequency of the spectrum of all initial samples in the training dataset, and the standard deviation of the frequency is the standard deviation of the main peak frequency of the spectrum of all initial samples in the training dataset. When the standard deviation of the frequency is a power of a preset value, the main peak frequency of the spectrum of each initial sample is scaled based on the average frequency and the standard deviation of the frequency to obtain the corresponding target sample main peak frequency. When the standard deviation of the frequency is not a power of a preset value, the frequency scaling factor is determined based on the standard deviation of the frequency, and the main peak frequency of the spectrum of each initial sample is scaled based on the average frequency and the frequency scaling factor to obtain the corresponding target sample main peak frequency. Furthermore, the average spectral amplitude and standard deviation of the maximum spectral amplitude of all initial samples in the training dataset are calculated. The average spectral amplitude is the average of the maximum spectral amplitudes of all initial samples in the training dataset, and the standard deviation is the standard deviation of the maximum spectral amplitudes of all initial samples in the training dataset. When the standard deviation of the spectral amplitude is a power of a preset value, the maximum spectral amplitude of each initial sample is scaled based on the average spectral amplitude and the standard deviation to obtain the corresponding maximum spectral amplitude of the target sample. When the standard deviation of the spectral amplitude is not a power of the preset value, a scaling factor is determined based on the standard deviation of the spectral amplitude, and the maximum spectral amplitude of each initial sample is scaled based on the average spectral amplitude and the scaling factor to obtain the corresponding maximum spectral amplitude of the target sample. For specific implementation details, please refer to the scaling details of the initial sample driving power described above.

[0053] Sb3. Combine the target sample driving power, target sample rotation speed fluctuation amplitude, and target sample frequency domain characteristics at the same time to form the corresponding target sample data.

[0054] Sb4. Input each target sample data into the initial eccentricity prediction model, use the corresponding eccentricity value label to guide the training output of the initial eccentricity prediction model, train the initial eccentricity prediction model, and obtain the intermediate eccentricity prediction model.

[0055] The initial eccentricity prediction model is an untrained neural network model framework used to learn the complex intrinsic relationship between drive power, speed fluctuation amplitude, and the frequency domain characteristics of speed and eccentricity mass, thereby achieving the eccentricity prediction function. The intermediate eccentricity prediction model is a model trained using multiple target sample data and their corresponding eccentricity value labels.

[0056] Specifically, each target sample data can be input into the initial eccentricity prediction model to obtain the corresponding training output. The initial eccentricity prediction model is then trained with the goal of minimizing the loss between the training output and the corresponding eccentricity value label to obtain the optimal eccentricity prediction model, i.e., the intermediate eccentricity prediction model.

[0057] Sb5. The intermediate model parameters of the intermediate eccentric prediction model are quantized to obtain the eccentric prediction model.

[0058] The intermediate eccentricity prediction model includes multiple network layers, each of which includes multiple intermediate weights and multiple intermediate biases. That is, the intermediate model parameters can include multiple intermediate weights and multiple intermediate biases. The target model parameters of the eccentricity prediction model can include multiple target weights and multiple target biases. Furthermore, the data type of the intermediate model parameters is floating point.

[0059] Specifically, a preset maximum quantization value and a preset minimum quantization value can be obtained. The preset maximum quantization value and the preset minimum quantization value are values ​​that are set in advance based on the data type of the target model parameter. That is, the preset maximum quantization value is the maximum value allowed within the range of the data type of the target model parameter, and the preset minimum quantization value is the minimum value allowed within the range of the data type of the target model parameter. For example, if it is known that the intermediate model parameter of floating point type needs to be quantized to int8 type, that is, the data type of the target model parameter is int8 type, then the preset maximum quantization value is 127 and the preset minimum quantization value is -128.

[0060] Next, any one of the multiple network layers in the intermediate eccentricity prediction model is selected as the current network layer. The maximum and minimum weights are determined from the multiple intermediate weights belonging to the current network layer. Then, based on the maximum weight, minimum weight, preset maximum quantization value, and preset minimum quantization value, the weight quantization factor of the current network layer is determined. This weight quantization factor characterizes the scaling factor used when performing integer quantization on the intermediate weights. Specifically, the product of the maximum weight and a preset value raised to a first objective power is set to equal the preset maximum quantization value, and the first objective power is calculated. The first objective power is then rounded to an integer. Next, the product of the minimum weight and a preset value raised to a second objective power is set to equal the preset minimum quantization value, and the second objective power is calculated. The second objective power is also rounded to an integer. If the first and second objective powers are equal, the first objective power is set as the weight quantization factor of the current network layer; otherwise, the minimum of the first and second objective powers is set as the weight quantization factor of the current network layer.

[0061] For example, if the data type of the target model parameters is known to be int8, the preset maximum quantization value is 127, the preset minimum quantization value is -128, and the preset value is 2, then the maximum weight of the current network layer is multiplied by 2. s1 =127, minimum weight of the current network layer ×2 s2 =﹣128, where s1 is the first objective power and s2 is the second objective power. The first objective power and the second objective power are solved to determine the weight quantization factor of the current network layer.

[0062] Next, based on the weight quantization factor of the current network layer, each intermediate weight belonging to the current network layer is integer-quantized to obtain the corresponding target weight. The target weight obtained at this time belongs to the current network layer in the eccentric prediction model; that is, the integer quantization formula can be: Where q is the target weight, f is the intermediate weight belonging to the current network layer, s is the weight quantization factor of the current network layer, m is the minimum quantization value, and n is the maximum quantization value. This represents the rounding operation, which maps the intermediate weights of each floating-point type to the target weights of the corresponding integer type. This can improve computational efficiency, reduce implementation complexity, and thus improve the efficiency and accuracy of determining the target weights.

[0063] Similarly, the maximum and minimum biases are determined from multiple intermediate biases belonging to the current network layer. Then, the bias quantization factor of the current network layer is determined based on the maximum bias, minimum bias, preset maximum quantization value, and preset minimum quantization value. Based on the bias quantization factor of the current network layer, each intermediate bias belonging to the current network layer is subjected to integer quantization to obtain the corresponding target bias. The integer quantization steps of the intermediate biases are the same as those of the intermediate weights. For the implementation details of the integer quantization steps of the intermediate weights, please refer to the above. It will not be repeated here.

[0064] Then, other network layers of the intermediate eccentric prediction model can be selected as the current network layer, and the above integer quantization steps can be repeated to obtain the target model parameters of the eccentric prediction model.

[0065] It should be noted that the intermediate eccentric prediction model includes the same network layers as the eccentric prediction model; each network layer corresponds to a weight quantization factor and a bias quantization factor.

[0066] In this embodiment, the initial eccentricity prediction model is trained with a large amount of sample data, enabling the intermediate eccentricity prediction model to capture the complex intrinsic relationship between drive power, speed fluctuation amplitude, and the frequency domain characteristics of speed and eccentricity mass, as well as learn feature patterns under various complex conditions. This improves the prediction accuracy of the intermediate prediction model and provides a reliable guarantee for the stable operation of the garment processing equipment. Furthermore, integer quantization is applied to the intermediate model parameters of the intermediate eccentricity prediction model, significantly reducing the memory space occupied by the model. This allows the model to run efficiently on processors with limited memory, fully utilizing hardware resources and avoiding operational errors or performance degradation due to insufficient memory, thereby improving the stability and reliability of the garment processing equipment. In addition, integer operations are faster, increasing the prediction speed of the eccentricity prediction model to the microsecond level, meeting the real-time control requirements of the garment processing equipment. This greatly improves the operating efficiency of the garment processing equipment, shortens garment processing time, and better meets the needs of modern users for an efficient lifestyle.

[0067] Alternatively, in another implementation, the feature selection steps are as follows: multiple candidate features can be randomly combined to form multiple input feature combinations. Then, any one of the input feature combinations is selected as the current input feature combination. The initial eccentricity prediction model is trained using the current input feature combination to obtain the intermediate eccentricity prediction model of the current input feature combination. If the loss decreases and tends to converge during the training process, it indicates that the model is effectively capturing the complex intrinsic relationship between input features and eccentricity quality. At this time, the candidate features included in the current input feature combination are determined as input features. If the loss does not decrease or continues to oscillate during the training process, it indicates that the input feature combination does not contain effective information.

[0068] Optionally, after training the intermediate eccentricity prediction model, any input feature can be selected as the current feature, and the data of the current feature in all the initial sample data included in the validation dataset can be randomly shuffled while keeping the data of other features unchanged. Then, the performance index of the intermediate eccentricity prediction model is evaluated based on the validation dataset. If the performance index drops by more than a preset drop value, it indicates that the current feature is strongly correlated with the eccentricity quality; otherwise, it indicates that the current feature is an invalid feature and is not correlated with the eccentricity quality.

[0069] S220: The current driving power, current speed fluctuation amplitude, and current frequency domain characteristics are scaled respectively to obtain the intermediate driving power, intermediate speed fluctuation amplitude, and intermediate frequency domain characteristics.

[0070] Among them, the intermediate drive power is the drive power obtained after scaling the current drive power, and the value range is [-1,1]; the intermediate speed fluctuation amplitude is the speed fluctuation amplitude obtained after scaling the current speed fluctuation amplitude, and the value range is [-1,1]; the intermediate frequency domain feature is the frequency domain feature obtained after scaling the current frequency domain feature, including the intermediate spectrum main peak frequency and the intermediate spectrum maximum amplitude, and the value range is [-1,1].

[0071] Specifically, when the power standard deviation is a power of a preset value, the current drive power is scaled based on the power average and the power standard deviation to obtain an intermediate drive power; when the power standard deviation is not a power of a preset value, the power scaling factor is determined based on the power standard deviation, and the current drive power is scaled based on the power average and the power scaling factor to obtain an intermediate drive power.

[0072] Meanwhile, when the standard deviation of the fluctuation amplitude is a power of a preset value, the current speed fluctuation amplitude is scaled based on the average fluctuation amplitude and the standard deviation of the fluctuation amplitude to obtain the intermediate speed fluctuation amplitude; when the standard deviation of the fluctuation amplitude is not a power of the preset value, the fluctuation amplitude scaling factor is determined based on the standard deviation of the fluctuation amplitude, and the current speed fluctuation amplitude is scaled based on the average fluctuation amplitude and the fluctuation amplitude scaling factor to obtain the intermediate speed fluctuation amplitude.

[0073] Similarly, when the frequency standard deviation is a power of a preset value, the current peak frequency of the spectrum is scaled based on the average frequency and the frequency standard deviation to obtain the intermediate peak frequency of the spectrum. When the frequency standard deviation is not a power of the preset value, a frequency scaling factor is determined based on the frequency standard deviation, and the current peak frequency of the spectrum is scaled based on the average frequency and the frequency scaling factor to obtain the intermediate peak frequency of the spectrum. Likewise, when the standard deviation of the spectral amplitude is a power of a preset value, the current maximum amplitude of the spectrum is scaled based on the average spectral amplitude and the spectral amplitude standard deviation to obtain the intermediate maximum amplitude of the spectrum. When the standard deviation of the spectral amplitude is not a power of the preset value, a spectral amplitude scaling factor is determined based on the spectral amplitude standard deviation, and the current maximum amplitude of the spectrum is scaled based on the average spectral amplitude and the spectral amplitude scaling factor to obtain the intermediate maximum amplitude of the spectrum.

[0074] For details on the specific implementation of the scaling process described above, please refer to the scaling process details of the initial sample drive power in Sb2, which will not be repeated here.

[0075] S230. Integer quantization is performed on the intermediate drive power, intermediate speed fluctuation amplitude and intermediate frequency domain characteristics respectively to obtain the target drive power, target speed fluctuation amplitude and target frequency domain characteristics.

[0076] Among them, the target driving power is the driving power obtained after integer quantization of the intermediate driving power; the target speed fluctuation amplitude is the speed fluctuation amplitude obtained after integer quantization of the intermediate speed fluctuation amplitude; the target frequency domain feature is the frequency domain feature obtained after integer quantization of the intermediate frequency domain feature, including the target spectrum main peak frequency and the target spectrum maximum amplitude.

[0077] Specifically, the maximum and minimum driving power are determined from all the initial sample driving powers included in the training dataset. Then, a power quantization factor is determined based on the maximum driving power, the minimum driving power, the preset maximum quantization value, and the preset minimum quantization value. The intermediate driving power is then integer quantized based on the power quantization factor to obtain the target driving power.

[0078] Meanwhile, the maximum and minimum speed fluctuation amplitudes are determined from all initial sample speed fluctuation amplitudes included in the training dataset. Then, the fluctuation amplitude quantization factor is determined based on the maximum speed fluctuation amplitude, the minimum speed fluctuation amplitude, the preset maximum quantization value, and the preset minimum quantization value. Based on the fluctuation amplitude quantization factor, the intermediate speed fluctuation amplitude is quantized by integer to obtain the target speed fluctuation amplitude.

[0079] Simultaneously, the maximum and minimum peak frequencies of the spectrum are determined from all initial sample peak frequencies included in the training dataset. Then, a frequency quantization factor is determined based on the maximum and minimum peak frequencies, a preset maximum quantization value, and a preset minimum quantization value. The intermediate peak frequencies are then subjected to integer quantization based on this frequency quantization factor to obtain the target peak frequency. Furthermore, the maximum and minimum maximum amplitude values ​​of the spectrum are determined from all initial sample maximum amplitude values ​​included in the training dataset. Then, a spectral amplitude quantization factor is determined based on the maximum and minimum maximum amplitude values, a preset maximum quantization value, and a preset minimum quantization value. The intermediate maximum amplitude values ​​are then subjected to integer quantization based on this spectral amplitude quantization factor to obtain the target maximum amplitude value.

[0080] For details on the specific implementation of the above integer quantization process, please refer to the details of the integer quantization process for intermediate weights in Sb5, which will not be repeated here.

[0081] S240. Using the eccentricity prediction model, the degree of eccentricity of the clothing processing equipment at the current moment is predicted based on the target driving power, the target speed fluctuation amplitude, and the target frequency domain characteristics, so as to obtain the target eccentricity prediction value.

[0082] Specifically, the target driving power, target speed fluctuation amplitude, and target frequency domain characteristics can be input into the eccentricity prediction model to obtain the initial eccentricity prediction value. That is, the target driving power, target speed fluctuation amplitude, target spectrum peak frequency, and target spectrum maximum amplitude can be input into the eccentricity prediction model. At this time, the eccentricity prediction model uses the target model parameters that have been learned to extract and analyze the features of the target driving power, target speed fluctuation amplitude, target spectrum peak frequency, and target spectrum maximum amplitude, and predicts the degree of eccentricity of the clothing processing equipment at the current moment, which is the initial eccentricity prediction value.

[0083] Since the initial eccentricity prediction value is determined based on the integer-type target model parameters and integer-type input data, its data type is integer. Therefore, after obtaining the initial eccentricity prediction value, it can be dequantized and descaled to obtain the target eccentricity prediction value. Specifically, the initial eccentricity prediction value can be dequantized to obtain the intermediate eccentricity prediction value. That is, f in the integer quantization formula of Sb5 can be considered as the target eccentricity prediction value, q as the initial eccentricity prediction value, and s as the eccentricity quantization factor, and the intermediate eccentricity prediction value can be obtained by solving. The data type of the intermediate eccentricity prediction value is then floating. Point type; then, the intermediate eccentricity prediction value is inversely scaled to obtain the target eccentricity prediction value. That is, the eccentricity mean and eccentricity standard deviation of all eccentricity value labels in the training dataset are calculated. When the eccentricity standard deviation is a power of a preset value, the intermediate eccentricity prediction value is inversely scaled based on the eccentricity mean and eccentricity standard deviation using the reverse scaling operation to obtain the target eccentricity prediction value. When the eccentricity standard deviation is not a power of a preset value, the eccentricity scaling coefficient is determined based on the eccentricity standard deviation, and the intermediate eccentricity prediction value is inversely scaled based on the eccentricity mean and eccentricity scaling coefficient using the reverse scaling operation to obtain the target eccentricity prediction value.

[0084] The steps for determining the eccentricity quantization factor are as follows: determine the maximum and minimum eccentricity values ​​among all eccentricity value labels included in the training dataset, and determine the eccentricity quantization factor based on the maximum eccentricity value, minimum eccentricity value, preset maximum quantization value, and preset minimum quantization value.

[0085] Optionally, the eccentric prediction model may include an input layer, a first hidden layer, a second hidden layer, and an output layer.

[0086] The input layer includes four neuron nodes, which are used to receive input data, namely the target driving power, the target rotational speed fluctuation amplitude, the target spectrum main peak frequency, and the target spectrum maximum amplitude.

[0087] The first hidden layer consists of 16 neurons. These 16 neurons are fully connected to the nodes in the input layer via a weight matrix. Each connection has a corresponding target weight, which measures the influence of the input data on that neuron. When the input data received from the input layer is passed to the neurons in the first hidden layer, each input data is multiplied by the target weight of that neuron and summed. Then, the target bias of that neuron is added, and a nonlinear transformation is performed through an activation function to obtain the output data of that neuron in the first hidden layer. Here, i represents the i-th neuron node in the input layer, and M represents the number of nodes in the input layer, i.e., the number of input data. This represents the input data received by the i-th neuron node in the input layer, and j represents the j-th neuron node in the first hidden layer. This represents the target weight corresponding to the connection between the i-th neuron node in the input layer and the j-th neuron node in the first hidden layer (i.e., the target weight of the j-th neuron node in the first hidden layer). This represents the target bias of the j-th neuron node in the first hidden layer. This represents the output data of the j-th neuron node in the first hidden layer; This represents the activation function, such as ReLU.

[0088] The second hidden layer consists of 8 neurons, which are fully connected to the nodes in the first hidden layer via a weight matrix. Each connection has a corresponding target weight, used to measure the influence of the output data from the first hidden layer on that neuron. When the output data from the first hidden layer is passed to the neurons in the second hidden layer, the output data of each neuron in the first hidden layer is multiplied by its target weight and summed. Then, the target bias of that neuron is added, and a nonlinear transformation is performed through an activation function to obtain the output data of that neuron in the second hidden layer. Here, i represents the i-th neuron node in the first hidden layer, and M represents the number of nodes in the first hidden layer. This represents the output data of the i-th neuron node in the first hidden layer, and j represents the j-th neuron node in the second hidden layer. This represents the target weight corresponding to the connection between the i-th neuron node in the first hidden layer and the j-th neuron node in the second hidden layer (i.e., the target weight of the j-th neuron node in the second hidden layer). This represents the target bias of the j-th neuron node in the second hidden layer. This represents the output data of the j-th neuron node in the second hidden layer.

[0089] The output layer consists of one neuron node, which is fully connected to nodes in the second hidden layer via a weight matrix. Each connection has a corresponding target weight, used to measure the influence of the output data from the second hidden layer on that neuron node. When the output data from the second hidden layer is passed to the neuron node in the output layer, the output data of each neuron node in the second hidden layer is multiplied by its target weight and summed. Then, the target bias of that neuron node is added, and a nonlinear transformation is performed through an activation function to obtain the initial eccentricity prediction value. Here, i represents the i-th neuron node in the second hidden layer, and M represents the number of nodes in the second hidden layer. This represents the output data of the i-th neuron node in the second hidden layer. denoted as the target weight (i.e., the target weight of the output layer neuron node) corresponding to the connection between the i-th neuron node in the second hidden layer and the neuron node in the output layer, b represents the target bias of the output layer neuron node, and Z represents the output data of the output layer neuron node, i.e., the initial bias prediction value.

[0090] For example, such as Figure 3 The diagram shown is a structural example of the eccentricity prediction model provided in an embodiment of this application. Figure 3 The target weight and target bias are not shown in the figure. Figure 3 In the diagram, each hidden layer's neurons are connected to the neurons in the previous layer, representing a fully connected relationship. While the target weights on these connections are not explicitly labeled, they reflect the weighted calculation process as data is transferred between neurons. Simultaneously, Figure 3 It also shows the data flow: the input data starts from the input layer, passes through the first hidden layer and the second hidden layer in sequence, and finally obtains the eccentricity prediction result, i.e. the initial eccentricity prediction value, in the output layer.

[0091] It should be noted that the input layer is only for receiving input data, and the neuron nodes therein do not involve model parameters such as weights and biases. Therefore, when the eccentricity prediction model includes an input layer, a first hidden layer, a second hidden layer, and an output layer, the eccentricity prediction model includes three network layers: a first hidden layer, a second hidden layer, and an output layer. Each of these network layers corresponds to a weight quantization factor and a bias quantization factor.

[0092] In this embodiment, two hidden layers are designed. The first hidden layer includes 16 neuron nodes, and the second hidden layer includes 8 neuron nodes. This design takes into account both the neural network's ability to fit complex data relationships and the resource limitations of the main processor of the clothing processing equipment control board. While ensuring a certain level of prediction accuracy, it minimizes the complexity and computational load of the eccentric prediction model.

[0093] S250. Determine the current eccentricity level based on the target eccentricity prediction value, and determine the balancing strategy that was pre-set for the current eccentricity level as the target balancing strategy.

[0094] Among them, the eccentricity level is used to characterize the degree of eccentricity of the garment processing equipment, and the higher the eccentricity level, the higher the degree of eccentricity; the current eccentricity level is the eccentricity level of the garment processing equipment at the current moment.

[0095] Optionally, the jitter parameters in the balancing strategy are positively correlated with the eccentricity level. The jitter parameters include jitter frequency, jitter amplitude, and jitter duration. That is, the higher the eccentricity level, the greater the jitter frequency, jitter amplitude, and jitter duration of the corresponding smoothing strategy.

[0096] Specifically, the current eccentricity level corresponding to the target eccentricity prediction value can be determined based on the relationship between the target eccentricity prediction value and the set eccentricity value, and the balancing strategy pre-set for the current eccentricity level can be determined as the target balancing strategy. Here, the set eccentricity value is a pre-set value used to determine the eccentricity level.

[0097] For example, the set eccentricity value includes a first set eccentricity value and a second set eccentricity value; when the target eccentricity prediction value is not greater than the first set eccentricity value, the current eccentricity level is determined as the first eccentricity level, and the target balance strategy is determined by controlling the drive motor to rotate forward and backward according to a first set frequency, a first set amplitude, and a first set duration, so that the load inside the garment processing equipment is redistributed under inertia; when the target eccentricity prediction value is greater than the first set eccentricity value but not greater than the second set eccentricity value, the current eccentricity level is determined as the second eccentricity level, and the target balance strategy is determined by controlling the drive motor to rotate forward and backward according to a second set frequency, a second set amplitude, and a second set duration, so that the load inside the garment processing equipment is redistributed under inertia; when the target eccentricity prediction value is greater than the second set eccentricity value, the current eccentricity level is determined as the third eccentricity level, and the target balance strategy is determined by controlling the drive motor to rotate forward and backward according to a third set frequency, a third set amplitude, and a third set duration, so that the load inside the garment processing equipment is redistributed under inertia.

[0098] S260, Execute the target balancing strategy to balance the load inside the garment processing equipment.

[0099] The technical solution of this application embodiment can collect the current driving power and current speed sequence of the clothing processing equipment, determine the current speed fluctuation amplitude based on the current speed sequence, and extract the current frequency domain features of the current speed sequence. Next, the current driving power, current speed fluctuation amplitude, and current frequency domain features are scaled to obtain intermediate driving power, intermediate speed fluctuation amplitude, and intermediate frequency domain features. This converts the data type of the input data to floating-point type, improving computational efficiency and reducing implementation complexity, thereby improving scaling efficiency and accuracy. Then, the intermediate driving power, intermediate speed fluctuation amplitude, and intermediate frequency domain features are quantized to obtain target driving power, target speed fluctuation amplitude, and target frequency domain features. This maps each floating-point data type to a corresponding integer data type, providing accurate input data for the subsequent eccentricity prediction model, improving computational efficiency and reducing implementation complexity, thereby improving the efficiency and accuracy of integer quantization. Finally, using the eccentricity prediction model, the degree of eccentricity of the clothing processing equipment at the current moment is predicted based on the target driving power, target speed fluctuation amplitude, and target frequency domain features, obtaining the target eccentricity prediction value. Utilizing the powerful nonlinear fitting capability of neural networks, a deep... This study explores the complex intrinsic relationship between drive power, speed fluctuation amplitude, frequency domain characteristics of speed, and eccentricity mass. Furthermore, by using integer data types for the target model parameters of the eccentricity prediction model, the computational load and memory usage of the model are significantly reduced, thereby improving prediction speed and meeting the real-time control requirements of garment processing equipment. This greatly enhances the operating efficiency of the garment processing equipment, shortens garment processing time, and better aligns with modern users' demands for efficient living. Subsequently, based on the target eccentricity prediction value, the current eccentricity level is determined, and the pre-set balancing strategy for the current eccentricity level is defined as the target balancing strategy. This ensures the target balancing strategy adapts to the actual eccentricity of the garment processing equipment, reducing unnecessary balancing operations caused by eccentricity prediction errors. Finally, the target balancing strategy is executed to balance the load within the garment processing equipment, effectively improving the eccentricity situation, reducing wear on key components, extending the equipment's lifespan, lowering maintenance costs, improving garment processing efficiency, protecting garments, and ensuring the safe, stable, and efficient operation of the garment processing equipment, ultimately enhancing the user experience.

[0100] Figure 4 This is a schematic diagram of the eccentric adjustment device of the clothing processing equipment provided in this application embodiment, referring to... Figure 4 The eccentric adjustment device of the garment processing equipment may include:

[0101] The acquisition model 410 is used to acquire the current driving power and current speed sequence of the clothing processing equipment, determine the current speed fluctuation amplitude based on the current speed sequence, and extract the current frequency domain features of the current speed sequence.

[0102] The prediction module 420 is used to predict the degree of eccentricity of the clothing processing equipment at the current moment based on the current driving power, the current speed fluctuation amplitude and the current frequency domain characteristics using an eccentricity prediction model, and to obtain the target eccentricity prediction value.

[0103] The execution module 430 is used to determine the target balancing strategy based on the target eccentricity prediction value and execute the target balancing strategy to balance the load inside the garment processing equipment.

[0104] In one embodiment, the data type of the target model parameter of the eccentricity prediction model is integer. The eccentricity adjustment device of the clothing processing equipment further includes a preprocessing module. The preprocessing module is used to perform scaling processing on the current driving power, current speed fluctuation amplitude, and current frequency domain features respectively before using the eccentricity prediction model to predict the degree of eccentricity of the clothing processing equipment at the current moment based on the current driving power, current speed fluctuation amplitude, and current frequency domain features to obtain intermediate driving power, intermediate speed fluctuation amplitude, and intermediate frequency domain features respectively; and to perform integer quantization processing on the intermediate driving power, intermediate speed fluctuation amplitude, and intermediate frequency domain features respectively to obtain target driving power, target speed fluctuation amplitude, and target frequency domain features respectively.

[0105] Accordingly, the prediction module 420 is specifically used to: use the eccentricity prediction model to predict the degree of eccentricity of the clothing processing equipment at the current moment based on the target driving power, the target speed fluctuation amplitude and the target frequency domain characteristics, and obtain the target eccentricity prediction value.

[0106] In one embodiment, the prediction module 420 is specifically used to: input the target driving power, the target speed fluctuation amplitude, and the target frequency domain characteristics into the eccentricity prediction model to obtain an initial eccentricity prediction value; and perform inverse quantization and inverse scaling on the initial eccentricity prediction value to obtain the target eccentricity prediction value.

[0107] In one embodiment, the training process of the eccentricity prediction model in the prediction module 420 is as follows: A training dataset is acquired, comprising multiple initial sample data and corresponding eccentricity value labels for each initial sample data. The initial sample data includes initial sample driving power, initial sample rotational speed fluctuation amplitude, and initial sample frequency domain features. The driving power of each initial sample is scaled to obtain the corresponding target sample driving power; the rotational speed fluctuation amplitude of each initial sample is scaled to obtain the corresponding target sample rotational speed fluctuation amplitude; and the frequency domain features of each initial sample are scaled to obtain the corresponding target sample frequency domain features. The target sample driving power, target sample rotational speed fluctuation amplitude, and target sample frequency domain features at the same time are combined to form the corresponding target sample data. Each target sample data is input into the initial eccentricity prediction model, and the corresponding eccentricity value label guides the training output of the initial eccentricity prediction model. The initial eccentricity prediction model is trained to obtain an intermediate eccentricity prediction model. The intermediate model parameters of the intermediate eccentricity prediction model are quantized to obtain the eccentricity prediction model.

[0108] In one embodiment, the prediction module 420 scales the driving power of each initial sample to obtain the corresponding target sample driving power, including: calculating the power average and power standard deviation of the driving power of all initial samples in the training dataset; when the power standard deviation is a power of a preset value, scaling the driving power of each initial sample based on the power average and power standard deviation to obtain the corresponding target sample driving power; when the power standard deviation is not a power of a preset value, determining the power scaling factor based on the power standard deviation, and scaling the driving power of each initial sample based on the power average and power scaling factor to obtain the corresponding target sample driving power.

[0109] In one embodiment, the intermediate model parameters include multiple intermediate weights and multiple intermediate biases, and the target model parameters of the eccentric prediction model include multiple target weights and multiple target biases. The prediction module 420 performs integer quantization processing on the intermediate model parameters of the intermediate eccentric prediction model to obtain the eccentric prediction model, including: selecting any one network layer from the multiple network layers included in the intermediate eccentric prediction model as the current network layer; determining the maximum weight and minimum weight from the multiple intermediate weights belonging to the current network layer; determining the weight quantization factor of the current network layer based on the maximum weight, minimum weight, preset maximum quantization value, and preset minimum quantization value; and performing integer quantization processing on each intermediate weight belonging to the current network layer based on the weight quantization factor of the current network layer to obtain the corresponding target weight; determining the maximum bias and minimum bias from the multiple intermediate biases belonging to the current network layer; determining the bias quantization factor of the current network layer based on the maximum bias, minimum bias, preset maximum quantization value, and preset minimum quantization value; and performing integer quantization processing on each intermediate bias belonging to the current network layer based on the bias quantization factor of the current network layer to obtain the corresponding target bias.

[0110] In one embodiment, the current frequency domain features include the current peak frequency of the spectrum and the current maximum amplitude of the spectrum. The input features of the eccentricity prediction model in the prediction module 420 are obtained by feature selection of multiple candidate features. The feature selection steps are as follows: determine the correlation between each candidate feature and the eccentricity quality to obtain the correlation of the corresponding candidate feature, and determine the importance between each candidate feature and the eccentricity quality to obtain the importance of the corresponding candidate feature; sort the multiple candidate features in descending order of correlation to obtain a first sorting result, and determine the relevant feature set based on the first sorting result and the top preset number of candidate features; sort the multiple candidate features in descending order of importance to obtain a second sorting result, and determine the important feature set based on the second sorting result and the top preset number of candidate features; and determine the intersection of the relevant feature set and the important feature set as the input feature.

[0111] In one embodiment, the execution module 430 determines the target balancing strategy based on the target eccentricity prediction value, including: determining the current eccentricity level based on the target eccentricity prediction value, and determining the balancing strategy pre-set for the current eccentricity level as the target balancing strategy; wherein, the jitter parameter in the balancing strategy is positively correlated with the eccentricity level.

[0112] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the above-described division of functional modules is merely an example. In practical applications, the above functions can be assigned to different functional modules as needed, that is, the internal structure of the device can be divided into different functional modules to complete all or part of the functions described above. The specific working process of the functional modules described above can be referred to the corresponding process in the foregoing method embodiments, and will not be repeated here.

[0113] The eccentricity adjustment device for the garment processing equipment provided in this embodiment can be applied to the eccentricity adjustment method for the garment processing equipment provided in any of the above embodiments, and has corresponding functions and beneficial effects.

[0114] Figure 5 This is a schematic diagram of the clothing processing device provided in an embodiment of this application. Figure 5 A block diagram is shown of an exemplary garment processing apparatus 11 suitable for implementing embodiments of the present application. Figure 5 The garment processing device 11 shown is merely an example and should not be construed as limiting the functionality and scope of use of this embodiment.

[0115] like Figure 5 As shown, the garment processing device 11 is presented in the form of a general-purpose computing electronic device. The components of the garment processing device 11 may include, but are not limited to: one or more processors or processing units 16, system memory 28, and bus 18 connecting different system components (including system memory 28 and processing unit 16).

[0116] Bus 18 represents one or more of several bus architectures, including a memory bus or memory controller, a peripheral bus, a graphics acceleration port, a processor, or a local bus using any of the various bus architectures. Examples of these architectures include, but are not limited to, industry-standard architecture buses, microchannel architecture buses, enhanced industry-standard architecture buses, Video Electronics Standards Association (VESA) local buses, and peripheral component interconnect buses.

[0117] The garment handling apparatus 11 typically includes a variety of computer-readable media. These media can be any available media that can be accessed by the garment handling apparatus 11, including volatile and non-volatile media, removable and non-removable media.

[0118] System memory 28 may include computer system readable media in the form of volatile memory, such as random access memory 30 and / or cache memory 32. The garment handling device 11 may further include other removable / non-removable, volatile / non-volatile computer system storage media. By way of example only, storage system 34 may be used to read and write non-removable, non-volatile magnetic media (…). Figure 5Not shown; usually referred to as a "hard drive"). Although Figure 5 As not shown, a disk drive for reading and writing to a removable non-volatile disk (e.g., a "floppy disk") and an optical disk drive for reading and writing to a removable non-volatile optical disk may be provided. In these cases, each drive may be connected to bus 18 via one or more data media interfaces. System memory 28 may include at least one program product having a set (e.g., at least one) of program modules configured to perform the functions of the embodiments of this application.

[0119] A program / utility 40 having a set (at least one) of program modules 42 may be stored, for example, in system memory 28. Such program modules 42 include, but are not limited to, an operating system, one or more application programs, other program modules, and program data. Each or some combination of these examples may include an implementation of a network environment. Program modules 42 typically perform the functions and / or methods described in the embodiments of this application.

[0120] The garment handling device 11 can also communicate with one or more external devices 14 (e.g., keyboard, pointing device, display 24, etc.), and with one or more devices that enable a user to interact with the garment handling device 11, and / or with any device that enables the garment handling device 11 to communicate with one or more other computing devices (e.g., network card and modem, etc.). This communication can be performed via the input / output interface 22. Furthermore, the garment handling device 11 can also communicate with one or more networks (e.g., local area network, wide area network, and / or public network) via the network adapter 20.

[0121] like Figure 5 As shown, network adapter 20 communicates with other modules of garment handling device 11 via bus 18. It should be understood that, although... Figure 5 As not shown, other hardware and / or software modules may be used in conjunction with the garment processing device 11, including but not limited to: microcode, device drivers, redundant processing units, external disk drive arrays, tape drives, and data backup storage systems.

[0122] The processing unit 16 executes various functional applications and page displays by running programs stored in the system memory 28, such as implementing an eccentric adjustment method for a clothing processing device provided in any embodiment of this application.

[0123] This application provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements, for example, an eccentric adjustment method for a garment processing device provided in any embodiment of this application.

[0124] The computer storage medium of this embodiment can be any combination of one or more computer-readable media. The computer-readable medium can be a computer-readable signal medium or a computer-readable storage medium. For example, a computer-readable storage medium can be, but is not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination thereof. More specific examples of computer-readable storage media (a non-exhaustive list) include: an electrical connection having one or more wires, a portable computer disk, a hard disk, a random access memory, a read-only memory, an erasable programmable read-only memory, an optical fiber, a portable compact disk read-only memory, an optical storage device, a magnetic storage device, or any suitable combination thereof. In this document, a computer-readable storage medium can be any tangible medium containing or storing a program that can be used by or in conjunction with an instruction execution system, apparatus, or device.

[0125] Computer-readable signal media may include data signals propagated in baseband or as part of a carrier wave, carrying computer-readable program code. Such propagated data signals may take various forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination thereof. Computer-readable signal media may also be any computer-readable medium other than computer-readable storage media, capable of sending, propagating, or transmitting programs for use by or in connection with an instruction execution system, apparatus, or device.

[0126] Program code contained on a computer-readable medium may be transmitted using any suitable medium, including but not limited to: wireless, wire, optical fiber, radio frequency, etc., or any suitable combination thereof.

[0127] Computer program code for performing the operations of this application can be written in one or more programming languages ​​or a combination thereof. Programming languages ​​include object-oriented programming languages ​​as well as conventional procedural programming languages. The program code can be executed entirely on the user's computer, partially on the user's computer, as a standalone software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In cases involving remote computers, the remote computer can be connected to the user's computer via any type of network, including a local area network (LAN) or a wide area network (WAN), or it can be connected to an external computer (e.g., via the Internet using an Internet service provider).

[0128] Those skilled in the art will understand that the modules or steps described above in this application can be implemented using general-purpose computing devices. They can be centralized on a single computing device or distributed across a network of multiple computing devices. Optionally, they can be implemented using computer-executable program code, which can then be stored in a storage device for execution by a computing device. Alternatively, they can be fabricated as separate integrated circuit modules, or multiple modules or steps can be fabricated as a single integrated circuit module. Thus, this application is not limited to any particular combination of hardware and software.

[0129] Note that the above are merely preferred embodiments and the technical principles employed in this application. Those skilled in the art will understand that this application is not limited to the specific embodiments described herein, and various obvious changes, readjustments, and substitutions can be made without departing from the scope of protection of this application. Therefore, although this application has been described in detail through the above embodiments, this application is not limited to the above embodiments. Many other equivalent embodiments may be included without departing from the inventive concept of this application, and the scope of this application is determined by the scope of the appended claims.

Claims

1. A method for adjusting the eccentricity of a garment processing device, characterized in that, The method includes: The current drive power and current speed sequence of the clothing processing equipment are collected, the current speed fluctuation amplitude is determined based on the current speed sequence, and the current frequency domain features of the current speed sequence are extracted. Using an eccentricity prediction model, the degree of eccentricity of the clothing processing equipment at the current moment is predicted based on the current driving power, the current speed fluctuation amplitude, and the current frequency domain characteristics, to obtain the target eccentricity prediction value; Based on the predicted target eccentricity value, a target balancing strategy is determined and executed to balance the load inside the garment processing equipment.

2. The eccentricity adjustment method for the garment processing equipment according to claim 1, characterized in that, The target model parameters of the eccentricity prediction model are of integer type. Before using the eccentricity prediction model to predict the degree of eccentricity of the clothing processing equipment at the current moment based on the current driving power, the current speed fluctuation amplitude, and the current frequency domain characteristics to obtain the target eccentricity prediction value, the following steps are also included: The current driving power, the current speed fluctuation amplitude, and the current frequency domain characteristics are scaled respectively to obtain the intermediate driving power, intermediate speed fluctuation amplitude, and intermediate frequency domain characteristics. The intermediate drive power, the intermediate speed fluctuation amplitude, and the intermediate frequency domain characteristics are respectively subjected to integer quantization processing to obtain the target drive power, the target speed fluctuation amplitude, and the target frequency domain characteristics. Accordingly, the method of using an eccentricity prediction model to predict the degree of eccentricity of the clothing processing equipment at the current moment based on the current driving power, the current speed fluctuation amplitude, and the current frequency domain characteristics, to obtain a target eccentricity prediction value, includes: Using the aforementioned eccentricity prediction model, the degree of eccentricity of the clothing processing equipment at the current moment is predicted based on the target driving power, the target rotational speed fluctuation amplitude, and the target frequency domain characteristics, thereby obtaining the target eccentricity prediction value.

3. The eccentricity adjustment method for the garment processing equipment according to claim 2, characterized in that, The step of using the eccentricity prediction model to predict the degree of eccentricity of the clothing processing equipment at the current moment based on the target driving power, the target rotational speed fluctuation amplitude, and the target frequency domain characteristics, to obtain the target eccentricity prediction value, includes: The target driving power, the target speed fluctuation amplitude, and the target frequency domain characteristics are input into the eccentricity prediction model to obtain the initial eccentricity prediction value. The initial eccentricity prediction value is dequantized and descaled to obtain the target eccentricity prediction value.

4. The eccentricity adjustment method for the garment processing equipment according to claim 1, characterized in that, The training process of the eccentricity prediction model is as follows: Obtain a training dataset, which includes multiple initial sample data and the eccentricity value label corresponding to each initial sample data. The initial sample data includes the initial sample driving power, the initial sample rotation speed fluctuation amplitude, and the initial sample frequency domain features. The driving power of each initial sample is scaled to obtain the driving power of the corresponding target sample; the speed fluctuation amplitude of each initial sample is scaled to obtain the speed fluctuation amplitude of the corresponding target sample; and the frequency domain feature of each initial sample is scaled to obtain the frequency domain feature of the corresponding target sample. The target sample driving power, target sample rotation speed fluctuation amplitude and target sample frequency domain characteristics at the same time are combined into the corresponding target sample data; Each target sample data is input into the initial eccentricity prediction model, and the corresponding eccentricity value label guides the training output of the initial eccentricity prediction model. The initial eccentricity prediction model is trained to obtain the intermediate eccentricity prediction model. The intermediate model parameters of the intermediate eccentricity prediction model are quantized to obtain the eccentricity prediction model.

5. The eccentricity adjustment method for the garment processing equipment according to claim 4, characterized in that, The scaling process for each initial sample driving power to obtain the corresponding target sample driving power includes: Calculate the power mean and power standard deviation of the driving power for all initial samples in the training dataset; When the power standard deviation is a power of a preset value, the driving power of each initial sample is scaled based on the power average and the power standard deviation to obtain the corresponding target sample driving power. When the power standard deviation is not a power of a preset value, a power scaling factor is determined based on the power standard deviation, and the driving power of each initial sample is scaled based on the power average value and the power scaling factor to obtain the corresponding target sample driving power.

6. The eccentricity adjustment method for the garment processing equipment according to claim 4, characterized in that, The intermediate model parameters include multiple intermediate weights and multiple intermediate biases, and the target model parameters of the eccentricity prediction model include multiple target weights and multiple target biases. The step of performing integer quantization on the intermediate model parameters of the intermediate eccentricity prediction model to obtain the eccentricity prediction model includes: Select any one network layer from the multiple network layers included in the intermediate eccentricity prediction model as the current network layer, determine the maximum weight and minimum weight from the multiple intermediate weights belonging to the current network layer, determine the weight quantization factor of the current network layer based on the maximum weight, the minimum weight, the preset maximum quantization value and the preset minimum quantization value, and perform integer quantization processing on each intermediate weight belonging to the current network layer based on the weight quantization factor of the current network layer to obtain the corresponding target weight; The maximum and minimum biases are determined from multiple intermediate biases belonging to the current network layer. The bias quantization factor of the current network layer is determined based on the maximum bias, the minimum bias, the preset maximum quantization value, and the preset minimum quantization value. Each intermediate bias belonging to the current network layer is then subjected to integer quantization based on the bias quantization factor of the current network layer to obtain the corresponding target bias.

7. The eccentricity adjustment method for the garment processing equipment according to claim 1, characterized in that, The current frequency domain features include the current main peak frequency and the current maximum amplitude of the spectrum. The input features of the eccentricity prediction model are obtained by feature selection from multiple candidate features. The feature selection steps are as follows: Determine the correlation between each candidate feature and the eccentricity quality to obtain the correlation of the corresponding candidate feature, and determine the importance between each candidate feature and the eccentricity quality to obtain the importance of the corresponding candidate feature; The candidate features are sorted in descending order of relevance to obtain a first sorting result, and a set of relevant features is determined based on the first sorting result and the first preset number of candidate features. The candidate features are sorted in descending order of importance to obtain a second sorting result, and an important feature set is determined based on the first preset number of candidate features in the second sorting result. The intersection of the relevant feature set and the important feature set is determined as the input feature.

8. The eccentricity adjustment method for the garment processing equipment according to claim 1, characterized in that, The step of determining the target balance strategy based on the predicted target eccentricity includes: The current eccentricity level is determined based on the target eccentricity prediction value, and the balancing strategy pre-set for the current eccentricity level is determined as the target balancing strategy; wherein, the jitter parameter in the balancing strategy is positively correlated with the eccentricity level.

9. A garment processing device, characterized in that, The garment processing equipment includes: At least one processor; and A memory communicatively connected to the at least one processor; wherein, The memory stores a computer program executable by the at least one processor, which enables the at least one processor to perform the eccentricity adjustment method of the garment processing device according to any one of claims 1 to 8.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When executed by the processor, the program implements the eccentric adjustment method of the garment processing device as described in any one of claims 1 to 8.

Images