Abnormality detection method for fabric treatment apparatus, electronic device, and fabric treatment apparatus

Abstract

The embodiment of the application discloses an abnormality detection method of a fabric treatment device, an electronic device and the fabric treatment device, and belongs to the technical field of intelligent detection and fault diagnosis of household appliances. The abnormality detection method of the fabric treatment device comprises the following steps: acquiring sound feature data and vibration feature data of the fabric treatment device in real time when the fabric treatment device is running; comparing the sound feature data and the vibration feature data with normal feature distribution in a digital twin model to obtain a difference measurement index; dynamically adjusting a pre-set abnormality judgment threshold value in combination with a current working condition of the fabric treatment device; and determining whether the fabric treatment device is abnormal according to the difference measurement index and the adjusted abnormality judgment threshold value. The embodiment of the application not only improves the detection accuracy under diversified working conditions, but also ensures the real-time performance and reliability of the equipment state monitoring through the dynamic updating capability of the digital twin model, thereby effectively avoiding the occurrence of major faults and safety hazards.

Description

Technical Field

[0001] This application relates to the field of intelligent detection and fault diagnosis technology for home appliances, and more specifically, to an anomaly detection method for fabric processing equipment, an electronic device, and the fabric processing equipment. Background Technology

[0002] With the continuous improvement of people's living standards and the level of economic and social development, the application of smart home appliances is becoming increasingly widespread and popular. As an indispensable electrical appliance in the home, the monitoring of the operating status and fault early warning of fabric processing equipment has become an important direction for improving user experience and extending equipment life. During the operation of fabric processing equipment, abnormal noise is often an important signal of internal mechanical failures (such as bearing wear, imbalance, transmission mechanism failure, etc.). Effective detection and diagnosis of such equipment is of great significance for avoiding major failures and safety hazards.

[0003] However, current methods for detecting anomalies in fabric processing equipment have certain limitations. These limitations result in low accuracy under diverse operating conditions and a lack of dynamic modeling capabilities for equipment status, impacting the timeliness and reliability of fault warnings. Summary of the Invention

[0004] This application provides an anomaly detection method, electronic device, and fabric processing equipment for fabric processing equipment, in order to at least solve the technical problems of low anomaly detection accuracy, poor adaptability, and lack of dynamic modeling capability in fabric processing equipment.

[0005] According to a first aspect of the embodiments of this application, an anomaly detection method for a fabric processing device is provided, comprising:

[0006] Real-time acquisition of sound and vibration characteristic data of fabric processing equipment during operation;

[0007] The sound feature data and vibration feature data are compared with the corresponding normal feature distribution in the digital twin model to obtain the difference measurement index. The normal feature distribution refers to the feature distribution obtained by training the digital twin model based on the sound feature data and vibration feature data collected by the fabric processing equipment under fault-free operation.

[0008] Based on the current operating conditions of the fabric processing equipment, the pre-set abnormal judgment threshold is dynamically adjusted. The current operating conditions include the aging status of the fabric processing equipment and the operating environment.

[0009] The presence of abnormalities in the fabric processing equipment is determined based on the difference measurement index and the adjusted anomaly judgment threshold.

[0010] This solution utilizes high-sensitivity sensors to collect real-time sound and vibration signals from fabric processing equipment during operation, extracting multi-dimensional features to form sound and vibration feature data. This data is input into a digital twin model and compared with a pre-built normal feature distribution to calculate a difference metric. Simultaneously, considering the equipment's aging condition and environmental conditions, the anomaly detection threshold is dynamically adjusted, making anomaly detection more closely aligned with actual operating conditions. The difference metric reflects the deviation between the current equipment state and the normal state; combined with the dynamically adjusted anomaly detection threshold, accurate identification of abnormal equipment states is achieved. This method not only improves detection accuracy under diverse operating conditions but also ensures the real-time nature and reliability of equipment status monitoring through the dynamic updating capability of the digital twin model, effectively preventing major failures and safety hazards.

[0011] In conjunction with the first aspect, in an optional implementation of this application embodiment, sound feature data and vibration feature data are compared with the corresponding normal feature distribution in the digital twin model to obtain a difference measurement index, including:

[0012] Calculate the Euclidean distance and / or cosine similarity between the sound feature data, vibration feature data and the corresponding normal feature distributions in the digital twin model;

[0013] Determine the difference measurement index based on the calculation results;

[0014] The presence of anomalies in the fabric processing equipment is determined based on the difference measurement index and the adjusted anomaly judgment threshold, including:

[0015] If the difference measurement index exceeds the anomaly judgment threshold, the fabric processing equipment is determined to be in an abnormal state.

[0016] This scheme employs Euclidean distance and cosine similarity as core algorithms to measure the differences between sound and vibration feature data and normal feature distributions from two dimensions: spatial distance and directional angle. Euclidean distance quantifies the absolute deviation between feature points, while cosine similarity focuses on capturing the directional consistency of feature vectors. By combining the results of these two algorithms, a difference metric is generated to further improve the accuracy of anomaly detection. When the difference metric exceeds a preset anomaly judgment threshold, the system determines that the device is in an abnormal state and triggers a subsequent early warning mechanism. This method significantly improves the adaptability and robustness of detection by modeling feature distributions under various operating conditions.

[0017] In conjunction with the first aspect, in one optional implementation of the embodiments of this application, the equipment aging status includes the equipment operating load and historical operating data, and the environmental conditions include ambient temperature, ambient humidity, and ambient noise. Therefore, based on the current operating condition of the fabric processing equipment, the pre-set anomaly judgment threshold is dynamically adjusted, including:

[0018] The operating load of the equipment and the historical operating condition data of the equipment are input into a long short-term memory network to determine the aging degree of the fabric treatment equipment;

[0019] The ambient temperature, ambient humidity, and ambient noise conditions are input into an environmental interference convolutional neural network to determine the interference factors of the fabric processing equipment.

[0020] The anomaly detection threshold is dynamically adjusted based on the degree of aging and the interfering factors.

[0021] This solution quantifies the aging level of equipment by analyzing its operating load and historical data. Simultaneously, it assesses the impact of external factors on equipment operation based on ambient temperature, humidity, and noise levels. This data is used to dynamically adjust anomaly detection thresholds to better reflect actual operating conditions. For example, when equipment aging is high or environmental interference is significant, the anomaly detection threshold is appropriately relaxed to avoid false alarms; conversely, when equipment is in good condition and the environment is stable, the threshold is tightened to improve detection sensitivity. This process significantly enhances the reliability and adaptability of anomaly detection.

[0022] In conjunction with the first aspect, in one optional implementation of the embodiments of this application, real-time acquisition of sound characteristic data and vibration characteristic data of the fabric processing equipment during operation includes:

[0023] The high-sensitivity acoustic and vibration sensors installed inside the washing machine collect sound and vibration signals in real time during operation.

[0024] The sound and vibration signals are preprocessed and multi-dimensional features are extracted to obtain sound feature data and vibration feature data.

[0025] In this solution, high-sensitivity acoustic and vibration sensors are fixedly installed in key areas of the fabric processing equipment, such as the motor housing, transmission mechanism, and roller support, to ensure comprehensive and accurate signal acquisition. After preprocessing steps such as filtering and noise reduction, the acquired sound and vibration signals are further extracted for multi-dimensional features, including spectral characteristics, temporal statistical characteristics, and energy distribution characteristics. These features collectively constitute sound and vibration feature data, providing a rich information foundation for subsequent anomaly detection. Through precise signal acquisition and feature extraction, the system's sensing capability and detection accuracy are significantly enhanced.

[0026] In conjunction with the first aspect, in an optional implementation of the embodiments of this application, the method further includes:

[0027] During the operation of the fabric processing equipment, the parameters of the digital twin model are dynamically updated through a sliding window mechanism to ensure that the model remains consistent with the actual equipment status.

[0028] This solution employs a sliding window mechanism to monitor and analyze the operational data of the fabric processing equipment in real time. The sliding window mechanism collects equipment operation data at fixed time intervals and inputs it into a digital twin model to update model parameters. For example, after the equipment has been running for a period of time, the wear and tear on its internal mechanical components may change, causing a shift in feature distribution. Through the sliding window mechanism, the system can promptly capture these changes and dynamically adjust the model parameters, ensuring that the model always reflects the actual state of the equipment. This continuous updating mechanism not only improves the model's adaptability but also enhances the reliability and timeliness of anomaly detection.

[0029] In conjunction with the first aspect, in one optional implementation of the embodiments of this application, the training method for the digital twin model includes:

[0030] The normal operation data of the fabric processing equipment under various working conditions were used as the training set, and the Mel frequency cepstral coefficient features and vibration time-domain statistical features of each frame signal were extracted to construct the initial feature space.

[0031] The K-means clustering algorithm is used to cluster the feature space and extract feature centers that represent the normal operating state;

[0032] The feature centers are trained using a support vector machine classifier to distinguish between normal and abnormal operating conditions, thus obtaining the normal feature distribution.

[0033] This scheme first collects normal operating data of the fabric processing equipment under different working conditions, such as no-load operation, full-load operation, and operation data under different load conditions. Then, it extracts Mel-frequency cepstral coefficient features and vibration time-domain statistical features from each frame of signal to form an initial feature space. The feature space is then clustered using the K-means clustering algorithm to extract feature centers that represent the normal operating state. Finally, a support vector machine classifier is used to train the feature centers to construct a normal feature distribution. This method, by modeling data under multiple working conditions, ensures the comprehensiveness and representativeness of the digital twin model, providing a reliable benchmark for anomaly detection.

[0034] In conjunction with the first aspect, in an optional implementation of this application embodiment, where the digital twin model is deployed on a cloud server, the method further includes:

[0035] Multi-device collaborative training and model sharing are achieved through cloud servers;

[0036] The performance of the digital twin model is analyzed and optimized through cloud servers, and the optimized model parameters are periodically retrieved from the cloud servers and deployed to the fabric processing equipment.

[0037] This solution deploys a digital twin model on a cloud server, leveraging the powerful computing capabilities of the cloud for collaborative training across multiple devices. Specifically, operational data from multiple fabric processing devices is uploaded to the cloud, and model parameters are optimized through joint training to improve the model's generalization ability. Simultaneously, the cloud server periodically evaluates and optimizes model performance, distributing the optimized model parameters to each device terminal to ensure the model on each device remains up-to-date. This cloud-terminal collaborative mechanism not only enhances model performance but also achieves efficient resource utilization and reduces the computational burden on individual devices.

[0038] In conjunction with the first aspect, in one optional implementation of this application embodiment, the digital twin model is a simplified digital twin model constructed based on the sliding window statistical method. The corresponding normal feature distribution in the simplified digital twin model is used as the baseline feature. Then, the sound feature data and vibration feature data are compared with the corresponding normal feature distribution in the digital twin model to obtain a difference measurement index, including:

[0039] The percentage deviation is calculated using the baseline features and the sound feature data and vibration feature data respectively, and the percentage deviation is used as a measure of difference.

[0040] The above method calculates the percentage deviation between the baseline features and the acoustic and vibration feature data, providing a clear indication of the degree of deviation between the equipment's operating state and its normal state. This method of difference measurement is simple and efficient, suitable for real-time anomaly detection scenarios, and further improves detection efficiency.

[0041] In conjunction with the first aspect, in one optional implementation of the embodiments of this application, dynamically adjusting the anomaly judgment threshold based on the degree of aging and interference factors includes:

[0042] The anomaly detection threshold is adjusted based on the degree of aging and interfering factors using the following formula:

[0043] W = W1X1 + W2X2 + W3X3;

[0044] Where X1 is the degree of aging, X2 is the interference factor, X3 is the anomaly judgment threshold, W is the adjusted anomaly judgment threshold, W1, W2, and W3 are the weight parameters of the degree of aging, interference factor, and anomaly judgment threshold, respectively, and the sum of W1, W2, and W3 is 1.

[0045] This scheme constructs a feature vector using mathematical formulas, and then weights and combines aging degree, interference factors, and anomaly detection threshold to generate an adjusted anomaly detection threshold. This process, through adjusting the weight parameters, comprehensively considers aging degree and interference factors, ensuring the scientific and flexible dynamic adjustment of the anomaly detection threshold. The variables in the formula interact to jointly determine the final threshold size, thereby improving the adaptability of anomaly detection, especially its detection capability under complex working conditions.

[0046] According to a second aspect of the embodiments of this application, an anomaly detection device for a fabric processing equipment is provided, comprising:

[0047] The acquisition unit is used to acquire, in real time, the acoustic and vibration characteristic data of the fabric processing equipment during operation.

[0048] The unit is used to compare sound feature data, vibration feature data, and the corresponding normal feature distribution in the digital twin model to obtain a difference measurement index.

[0049] The processing unit is used to dynamically adjust the pre-set abnormal judgment threshold based on the current operating conditions of the fabric processing equipment.

[0050] The processing unit is also used to determine whether there is an anomaly in the fabric processing equipment based on the difference measurement index and the adjusted anomaly judgment threshold.

[0051] According to a third aspect of the embodiments of this application, the present invention provides an electronic device, including: a memory and a processor, the memory and the processor being communicatively connected to each other, the memory storing computer instructions, and the processor executing the computer instructions to perform the abnormal detection method of the fabric processing device described in the first aspect or any corresponding embodiment.

[0052] According to a fourth aspect of the embodiments of this application, this specification provides a computer-readable storage medium storing computer instructions that, when executed by a processor, implement the anomaly detection method for the fabric processing equipment as described in any of the preceding claims.

[0053] According to a fifth aspect of the embodiments of this application, this specification provides a computer program product or computer program, the computer program product including a computer program stored in a computer-readable storage medium; a processor of a computer device reads the computer program from the computer-readable storage medium, and when the processor executes the computer program, it implements the abnormal detection method of the fabric processing equipment as described in any of the preceding claims.

[0054] According to a sixth aspect of the embodiments of this application, this specification provides a fabric treatment apparatus that employs an anomaly detection method for fabric treatment apparatus as described in any of the first aspects, or has electronic equipment as described in the third aspect.

[0055] The technical effects achieved by the second to sixth aspects mentioned above are similar to those achieved by the corresponding technical means in the first aspect, and will not be repeated here. Attached Figure Description

[0056] Figure 1 This is a flowchart illustrating the anomaly detection method for the fabric processing equipment provided in the embodiments of this application;

[0057] Figure 2 This is a schematic diagram of the noise acquisition and feature extraction process provided in the embodiments of this application;

[0058] Figure 3 This application provides a schematic diagram of the anomaly detection and feedback logic flow in its embodiments.

[0059] Figure 4 This is a flowchart illustrating the digital twin model training method provided in an embodiment of this application;

[0060] Figure 5 This is a schematic diagram of the structure of the abnormality detection device of the fabric processing equipment provided in the embodiments of this application;

[0061] Figure 6 This is a schematic diagram of the structure of the electronic device provided in the embodiments of this application. Detailed Implementation

[0062] To enable those skilled in the art to better understand the present application, the technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present application, and not all embodiments. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative effort should fall within the scope of protection of the present application.

[0063] It should be understood that "multiple" as mentioned herein refers to two or more. In the description of the embodiments of this application, unless otherwise stated, " / " means "or," for example, A / B can mean A or B; "and / or" in this document is merely a description of the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A existing alone, A and B existing simultaneously, and B existing alone. In addition, to facilitate a clear description of the technical solutions of the embodiments of this application, the terms "first," "second," etc., are used in the embodiments of this application to distinguish identical or similar items with substantially the same function and effect. Those skilled in the art will understand that the terms "first," "second," etc., do not limit the quantity or execution order, and the terms "first," "second," etc., do not necessarily imply that they are different.

[0064] Furthermore, the terms “comprising” and “having”, and any variations thereof, are intended to cover non-exclusive inclusion, such that a process, method, system, product, or apparatus that includes a series of steps or units is not necessarily limited to those steps or units that are explicitly listed, but may include other steps or units that are not explicitly listed or that are inherent to such process, method, product, or apparatus.

[0065] As mentioned in the background section, with the continuous improvement of people's living standards and the level of economic and social development, the application of smart home appliances is becoming increasingly widespread and popular. Fabric processing equipment, as an indispensable household appliance, requires monitoring of its operational status and fault early warning systems, which has become an important direction for improving user experience and extending equipment lifespan. During operation, abnormal noise in fabric processing equipment is often a significant signal of internal mechanical failures (such as bearing wear, imbalance, and transmission mechanism malfunctions). Effective detection and diagnosis of such equipment are crucial for avoiding major failures and safety hazards.

[0066] However, current methods for detecting anomalies in fabric processing equipment have certain limitations. Manual listening involves judging the presence of abnormal sounds by listening to and observing the equipment's operating status, but this method relies on human experience and struggles to meet real-time and continuous requirements. Sensor-based signal acquisition and threshold alarm methods use microphones and vibration sensors to collect sound and vibration signals during operation and determine anomalies by setting fixed thresholds, but this method lacks adaptability to different operating conditions and is prone to false alarms or missed alarms. Traditional machine learning methods extract feature data to train classification models for fault identification, but their adaptability to changes in equipment operating conditions is limited, and they struggle to reflect dynamic changes in equipment status in real time. These problems result in low detection accuracy of existing technologies under diverse operating conditions and a lack of dynamic modeling capabilities for equipment status, affecting the timeliness and reliability of fault warnings.

[0067] Based on this, embodiments of this application provide an anomaly detection method for fabric processing equipment, such as... Figure 1 As shown, this invention provides an anomaly detection method for fabric processing equipment. Its core lies in acquiring sound and vibration signals from the equipment during operation using a high-sensitivity sensor, and analyzing these signals in conjunction with a digital twin model to achieve accurate identification of abnormal equipment states. The specific implementation process of this method will be described in detail below, including signal acquisition, feature extraction, difference measurement calculation, dynamic threshold adjustment, and the construction and updating of the digital twin model. The method includes the following processing steps.

[0068] S101: Real-time acquisition of sound and vibration characteristic data of the fabric processing equipment during operation.

[0069] In practice, acoustic and vibration sensors are installed in key locations within the fabric processing equipment, including the motor housing, transmission mechanism, and roller supports. Acoustic sensors collect sound signals generated during equipment operation, reflecting air vibrations. Vibration sensors collect vibration signals generated during equipment operation, reflecting structural vibrations. The sensor placement is carefully designed to comprehensively cover the main mechanical components of the equipment, ensuring comprehensive and accurate signal acquisition. For example, installing acoustic sensors on the motor housing captures the noise characteristics of the motor during operation, while installing vibration sensors near the transmission mechanism monitors the vibration of transmission components. To reduce the impact of environmental noise on signal acquisition, the sensors employ a high-sensitivity design and are equipped with shielding layers to improve the signal-to-noise ratio.

[0070] Then, the feature data undergoes preprocessing. Specifically, the collected sound and vibration signals need to be preprocessed before they can be used for subsequent analysis. Preprocessing mainly includes two stages: filtering and noise reduction. In the filtering stage, a bandpass filter is used to filter the signal by frequency, removing low-frequency interference and high-frequency noise, and retaining the frequency bands closely related to the equipment's operating status. For example, for fabric processing equipment, typical operating noise is mainly concentrated between 500Hz and 5kHz, so the passband range of the filter is set to this range. In the noise reduction stage, a wavelet transform algorithm is used to decompose and reconstruct the signal, further removing random noise components. After preprocessing, multi-dimensional feature extraction is performed on the signal, including spectral features, time-domain statistical features, and energy distribution features. Spectral features are calculated using a fast Fourier transform, reflecting the energy distribution of the signal at different frequencies; time-domain statistical features include parameters such as mean, variance, and peak factor, used to describe the signal's variation over time; energy distribution features are obtained by calculating the energy value of the signal in segments, reflecting the energy fluctuation of the signal over time. These features together constitute sound feature data and vibration feature data, and their execution process is as follows: Figure 2 As shown.

[0071] S102: Compare the sound feature data and vibration feature data with the corresponding normal feature distribution in the digital twin model to obtain the difference measurement index.

[0072] In practice, the extracted sound and vibration feature data are input into the digital twin model and compared with a pre-built normal feature distribution for analysis. For example... Figure 3 As shown, during the comparative analysis, the Euclidean distance and cosine similarity between the sound feature data, vibration feature data, and the normal feature distribution in the digital twin model are calculated. It should be noted that the normal feature distribution refers to the feature distribution obtained by training the digital twin model based on feature data and vibration feature data collected from the fabric processing equipment under fault-free operating conditions. Euclidean distance is used to quantify the absolute deviation between feature points, while cosine similarity focuses on capturing the directional consistency of feature vectors. In this step, either one can be chosen for comparison, or both can be used. When both are used, a difference metric can be generated by combining the results of these two algorithms. The formula for calculating the difference metric is D = αd + β(1 - cosθ), used to balance the contributions of the two algorithms, where α and β are weighting coefficients, and d and cosθ are the Euclidean distance and cosine similarity, respectively. When the difference metric exceeds a preset anomaly judgment threshold, the system determines that the equipment is in an abnormal state.

[0073] The digital twin model referred to in this application is a comprehensive virtual entity integrating a physical model, a behavioral model, and a data model. It receives real-time acoustic and vibration characteristic data, as well as equipment operating condition data, from the fabric processing equipment. The model mainly includes:

[0074] 1. Physical-behavioral model: Based on the equipment's CAD / CAE data and operating mechanism, construct the equipment's virtual geometry and the dynamic model of key components to simulate its ideal acoustic and vibration behavior under different working conditions.

[0075] 2. Data-driven model: Learn the distribution of normal operating characteristics and abnormal patterns of the device from historical and real-time sensor data through deep learning networks (such as LSTM for aging prediction and CNN for environmental interference identification).

[0076] 3. Decision and Feedback Module: This module compares real-time data with the normal feature distribution within the model, calculates a difference metric, and uses dynamically adjusted thresholds to determine anomalies. It also provides early warnings and maintenance suggestions to actual equipment or users based on the model's predictions.

[0077] In another example, for a feature vector X collected at a certain moment and the center vector Y of the normal feature distribution, the difference metric D can be determined by the following formula:

[0078] The calculations show that X1 to Xn and Y1 to Yn are the values ​​of each dimension of the feature vector and the center vector, respectively. The magnitude of the difference metric reflects the degree of deviation between the current operating state and the normal state.

[0079] S103: Based on the current operating conditions of the fabric processing equipment, dynamically adjust the pre-set abnormal judgment threshold.

[0080] In practice, the current operating conditions include the aging status of the fabric processing equipment and the environmental conditions. To improve the flexibility and accuracy of anomaly detection in this step, the anomaly judgment threshold needs to be dynamically adjusted based on the equipment's aging status and the environmental conditions before comparing feature data. The aging status is assessed using information such as the equipment's cumulative operating time and historical maintenance records. An aging coefficient is introduced to correct the threshold, making it more consistent with the current actual condition of the equipment. Environmental conditions include external factors such as temperature and humidity, which are monitored in real time by environmental sensors and fed back to the system. The system dynamically adjusts the threshold based on changes in environmental parameters; for example, the threshold is appropriately relaxed in high-temperature environments to avoid false positives. This dynamic adjustment mechanism ensures the adaptability of anomaly detection.

[0081] In one possible implementation, the aging condition of equipment can be quantified using the cumulative value of the equipment's operating load and historical operating data. Specifically, the cumulative value of the equipment's operating load reflects the degree of fatigue experienced by the equipment during long-term operation, and is calculated by integrating the load values ​​of the equipment over different time periods. Historical operating data includes the number of equipment failures and maintenance records, which reflect the equipment's reliability level. The degree of equipment aging can be obtained by weighted summation of the fatigue coefficient and reliability coefficient. Environmental conditions include ambient temperature, humidity, and noise levels. The impact of ambient temperature and humidity can be quantified by calculating their deviation from the equipment's rated operating conditions; the greater the deviation, the higher the interference factor. The impact of ambient noise is assessed by comparing its difference from the equipment's rated noise level. Finally, the temperature interference factor, humidity interference factor, and noise interference factor are weighted and summed to obtain the comprehensive interference factor. Based on the degree of equipment aging and the comprehensive interference factor, the anomaly judgment threshold is dynamically adjusted to better reflect the actual operating conditions of the equipment.

[0082] In one possible implementation, a mechanism for dynamically adjusting the anomaly detection threshold needs to be designed to address the aging condition of the equipment and the operating environment. The aging condition includes the equipment's operating load and historical operating data, such as cumulative operating time and historical fault records. The operating environment encompasses ambient temperature, humidity, and noise levels. In implementation, ambient temperature, humidity, and noise data are first collected using temperature and humidity sensors and a noise monitor, and this data is then input into an environmental interference convolutional neural network. This network consists of multiple convolutional and fully connected layers, which extract features from the input data and perform nonlinear mapping to output a value representing the intensity of environmental interference. Simultaneously, the aging degree of the equipment is calculated using a linear regression model based on its historical operating data. The aging degree and interference factors are then substituted into the formula W = W1X1 + W2X2 + W3X3, where X1 represents the aging degree, X2 represents the interference factor, X3 represents the initial anomaly detection threshold, and W1, W2, and W3 are the weighting coefficients for the corresponding parameters. By adjusting these weighting coefficients, the anomaly detection threshold can be dynamically adjusted to adapt to the detection needs under different operating conditions. During the aging-dominant phase, W1 can be appropriately increased (0.5-0.7) and W3 decreased (0.2-0.3). During the disturbance emergence phase, W2 can be increased (0.4-0.6) and W1 appropriately decreased. During normal operation, the baseline weights are maintained.

[0083] In another example, a Long Short-Term Memory (LSTM) network is used to model the aging process of equipment. The input to the LSTM network is a time series of data consisting of statistical values ​​and operating parameters of the equipment's operating characteristics collected at fixed intervals (e.g., every 24 hours or every 10 operating cycles). These statistics include: the average L2 norm of the acoustic MFCC features over the past 24 hours, the mean of the vibration RMS values, the average rate of change of the FFT dominant frequency amplitude, and the average load exponent. The LSTM network consists of an input layer, two hidden LSTM layers (each with 128 units), and a fully connected output layer. The LSTM layers effectively capture long-term dependencies through their internal gating mechanisms (input gate, forget gate, output gate), learning and predicting the wear, fatigue, and performance degradation states of the equipment. For example, bearing wear typically manifests as a slow increase in high-frequency components in the vibration signal, or an amplitude drift of specific harmonic components; motor carbon brush wear may lead to changes in the brush noise spectrum. The LSTM network predicts the degree of equipment aging by analyzing these progressive characteristic changes. The fully connected output layer outputs a normalized, continuous aging index, for example, between 0 and 1, where 0 represents a brand-new device and 1 represents reaching the end of its designed lifespan. The LSTM network is trained using backpropagation and optimizes a mean squared error loss function to predict an aging index that matches actual aging levels (e.g., data from physical inspection, disassembly assessments, or accelerated life testing). The optimizer uses Adam with a learning rate of 0.001 and a batch size of 32.

[0084] An environmental noise convolutional neural network (CNN) is used to identify and quantify the intensity of ambient background noise. The CNN input is a Mel spectrogram of the ambient background noise, obtained by processing the background noise acquired by a microphone using a short-time Fourier transform (STFT) and a Mel filter bank; its dimension is 128x128. The CNN network contains three convolutional layers, each using a 3x3 kernel, a ReLU activation function, and batch normalization. The first layer contains 32 kernels, the second 64, and the third 128. Each convolutional layer is followed by a 2x2 max-pooling layer to extract multi-scale features and reduce dimensionality. This is followed by a flattening layer connected to a fully connected layer with 256 neurons, ultimately outputting an environmental noise factor through a sigmoid activation function. This intensity factor ranges from, for example, 0 to 1, where 0 represents no ambient noise interference and 1 represents strong ambient noise interference. CNNs learn specific patterns in the spectrogram associated with common environmental noise sources (e.g., human voices, television sounds, other appliance noises, fan noise) to distinguish them from the device's own operating noise and estimate the masking effect of environmental noise. For example, this factor increases when a significant low-frequency environmental hum is detected. Temperature and humidity data (normalized) are processed by a separate small multilayer perceptron (MLP), whose output is multiplied or added to the CNN output to form the final environmental interference factor, comprehensively characterizing the impact of the environment on sensor data acquisition and device operating characteristics.

[0085] Subsequently, traditional fixed-threshold anomaly detection methods often exhibit high false alarm and false negative rates when faced with dynamically changing operating conditions, gradual aging, and complex and variable environmental disturbances in fabric processing equipment. This invention introduces a dynamic adjustment mechanism for the anomaly judgment threshold to adapt to the complexity of equipment operation. The dynamic adjustment process is implemented through a Threshold Adaptation Network (TAN). TAN is a specially trained multilayer perceptron (MLP), whose inputs include a load index, an aging index, an environmental disturbance factor, and a preset empirical baseline threshold. The baseline threshold can be set to an initial value, such as 10.0, based on the equipment type and historical experience.

[0086] The TAN network structure consists of an input layer (4 neurons, corresponding to the load index, aging index, environmental interference factor, and baseline threshold), and three hidden layers, each containing 256 neurons with ReLU activation function. Batch normalization is used between the hidden layers to accelerate training and improve stability. The final output layer is a linear activation layer with a single neuron, producing a scalar value, which is the anomaly detection threshold dynamically adjusted according to the current operating conditions.

[0087] TAN's training data includes a large amount of historical operational data, encompassing the characteristic fluctuation ranges under different operating parameters during normal operation, as well as the characteristic behavior of confirmed abnormal events under different operating conditions. For example, under heavy load conditions (load index 0.8-1.0), the amplitude fluctuations of normal vibration and acoustic signals are large. In this case, TAN will output a relatively high threshold (e.g., 15.0) to avoid false alarms caused by normal fluctuations. In the early fault stage under light load (load index 0.2-0.4, aging index 0.1-0.3), the abnormal equipment signals may be very weak. In this case, TAN will output a relatively sensitive threshold (e.g., 8.0) to ensure that early anomalies can be captured in a timely manner. When significant environmental noise (environmental interference factor 0.6-1.0) is detected, the threshold will be adjusted upwards accordingly (e.g., 12.0) to compensate for the superimposed effect of noise on sensor data.

[0088] In a preferred embodiment of the present invention, the training process of the Threshold Adaptive Network (TAN) utilizes a reinforcement learning (RL) strategy to learn the optimal threshold adjustment strategy. The RL agent takes feature vectors and operating parameters (including difference metrics, load indices, aging indices, and environmental interference factors) from historical operating data as state inputs and uses different threshold adjustment strategies (e.g., fine-tuning the threshold increment or multiplicative factor) as actions. The reward signal is a combined metric function based on the false positive rate, false negative rate, and anomaly detection delay. Specifically, the reward function can be designed as follows: Where FPR is the false alarm rate, FNR is the false negative rate, and Latency is the detection latency. These are the weighting coefficients. Through extensive trial-and-error learning in simulated environments (e.g., simulating equipment operation and failures on offline datasets), TAN learns an optimal threshold adjustment strategy that minimizes false positives and false negatives while ensuring sensitivity for early anomaly detection under various complex operating conditions. The RL agent employs a Deep Q-Network (DQN) architecture, storing and reusing historical experience through an Experience Replay mechanism, and improving training stability and convergence speed through a Target Network. The neural network structure of DQN is similar to the MLP described above, but its output layer corresponds to the Q-values ​​of different actions.

[0089] The digital twin model in this solution can be deployed locally or in the cloud; this disclosure does not limit this. If deployed on a cloud server, the powerful computing capabilities of the cloud are utilized for multi-device collaborative training. Operational data from multiple fabric processing devices are uploaded to the cloud, and model parameters are optimized through joint training to improve the model's generalization ability. Simultaneously, the cloud server periodically evaluates and optimizes model performance, distributing the optimized model parameters to each device terminal to ensure the model on the device is always up-to-date. This cloud-terminal collaborative mechanism not only improves model performance but also achieves efficient resource utilization and reduces the computational burden on individual devices.

[0090] In this step, during the operation of the fabric processing equipment, the parameters of the digital twin model can be dynamically updated through a sliding window mechanism to ensure that the model remains consistent with the actual equipment state. The sliding window mechanism collects equipment operation data at fixed time intervals and inputs it into the digital twin model to update model parameters. For example, after the equipment has been running for a period of time, the wear and tear on its internal mechanical components may change, causing a shift in feature distribution. Through the sliding window mechanism, the system can promptly capture these changes and dynamically adjust the model parameters, ensuring that the model always reflects the actual state of the equipment. This continuous updating mechanism not only improves the model's adaptability but also enhances the reliability and timeliness of anomaly detection.

[0091] The parameters of the digital twin model are dynamically updated using a sliding window mechanism to ensure that the model remains consistent with the actual equipment state. Specifically, the sliding window periodically (e.g., every N operating cycles or M hours) collects the latest normal operation data to retrain or fine-tune the weights and biases of the data-driven model part of the digital twin model, as well as parameters such as feature centers of K-means clustering and decision boundaries of SVM classifiers. In addition, certain parameters in the physical-behavioral model (such as wear coefficient and stiffness parameters) can also be dynamically adjusted based on the assessment results of the equipment aging degree to more accurately reflect the current physical state of the equipment.

[0092] S104: Determine whether there is an anomaly in the fabric processing equipment based on the difference measurement index and the adjusted anomaly judgment threshold.

[0093] In practice, when the system detects an anomaly in the fabric processing equipment, it generates detailed alarm information, including the type and severity of the anomaly, as well as possible cause analysis. This information is pushed to the user in real time via the user interface or mobile application, reminding the user to take appropriate measures. For example, if abnormal wear is detected in the internal bearings of the equipment, the system will prompt the user to arrange repairs as soon as possible to prevent the fault from escalating further. Timely alarm feedback significantly improves the user experience and effectively extends the service life of the equipment.

[0094] The entire anomaly detection method encompasses every step from signal acquisition to anomaly determination, with each step meticulously designed and optimized to ensure the system's detection accuracy and reliability. Through precise signal acquisition by high-sensitivity sensors, comprehensive analysis via multi-dimensional feature extraction, intelligent comparison using digital twin models, and flexible dynamic threshold adjustment, this method can accurately identify abnormal states in fabric processing equipment under diverse operating conditions, effectively preventing major malfunctions and safety hazards.

[0095] This embodiment can be executed based on the anomaly detection system of the fabric processing equipment, which specifically includes the following main hardware modules:

[0096] ① Sensing Module: A high-sensitivity microphone and a triaxial vibration sensor installed inside the fabric processing equipment are responsible for collecting sound and vibration signals during the operation of the equipment. The microphone's frequency range covers 20 Hz to 20 kHz, and the vibration sensor collects vibration acceleration data with a sampling frequency of 1 kHz or higher.

[0097] ② Communication Module: Employs a low-power WiFi or Bluetooth module to enable wireless data transmission between the sensing module and the local gateway or cloud server. Supports data caching to address network instability.

[0098] ③ Data processing module: Deployed on a local server or in the cloud, it is responsible for signal preprocessing, feature extraction, construction and updating of digital twin models, and operation of anomaly detection algorithms.

[0099] ④ Interactive feedback module: Real-time feedback of abnormal alarm information to users via mobile app, SMS push or cloud platform interface, and supports remote fault diagnosis and repair requests.

[0100] like Figure 4 As shown, the digital twin model training method in this application embodiment is described, which specifically includes the following processing steps:

[0101] S401: Use the normal operation data of the fabric processing equipment under various working conditions as the training set, and extract the Mel frequency cepstral coefficient features and vibration time-domain statistical features of each frame signal to construct the initial feature space.

[0102] S402: The K-means clustering algorithm is used to cluster the feature space and extract feature centers that represent the normal operating state.

[0103] S403: Use a support vector machine classifier to train the feature centers, distinguish between normal and abnormal operating conditions, and obtain the normal feature distribution.

[0104] The construction process of the digital twin model is as follows: First, normal operation data of the fabric processing equipment under various working conditions is collected, including no-load operation, full-load operation, and operation data under different load conditions. Then, Mel frequency cepstral coefficient features and vibration time-domain statistical features are extracted from each frame of signal to form an initial feature space. The Mel frequency cepstral coefficient features are obtained by processing the signal through short-time Fourier transform and Mel filter bank, effectively characterizing the spectral characteristics of the sound signal; the vibration time-domain statistical features include parameters such as root mean square value, kurtosis, and skewness, used to describe the statistical characteristics of the vibration signal. Next, the K-means clustering algorithm is used to cluster the feature space, extracting feature centers that represent the normal operating state. Finally, a support vector machine classifier is used to train the feature centers to distinguish between normal and abnormal operating conditions, obtaining the normal feature distribution. This method of modeling based on multi-condition data ensures the comprehensiveness and representativeness of the digital twin model.

[0105] It should be noted that initially, the same model corresponds to the same fabric processing equipment. Subsequently, when it is installed in each piece of equipment or stored in the cloud in a one-to-one correspondence with each equipment, the model is adjusted according to the changes of each equipment.

[0106] In one example, the fabric handling equipment is a washing machine. First, the washing machine is set to a "standard wash" program. During operation, 10 minutes of sound and vibration data are simultaneously collected. A bandpass filter is then used to filter out low-frequency noise (0-20 Hz) and electromagnetic interference above 10 kHz. Z-score normalization is then applied to each channel for standardization. The signal is then divided into frames with a 20 ms frame length and a 10 ms frame shift, and weighted using a Hamming window. The aforementioned 10 minutes of normal operation data is then used as the training set. MFCC features (13 dimensions) and vibration time-domain statistical features (mean, standard deviation, kurtosis, skewness, dominant frequency) are extracted from each frame, totaling 20 dimensions. The K-means algorithm is then used to cluster the feature space, extracting feature centers representing normal operation as the initial state of the digital twin model. Finally, an SVM classifier is used as the backend detection model, trained to distinguish between "normal" and "abnormal" operating conditions.

[0107] Training for "abnormal" operating conditions can also be conducted in the following ways:

[0108] 1. Insert three metal nuts into the washing machine drum to simulate abnormal noise caused by a loose internal structure of the drum.

[0109] 2. The system collects signals and extracts features in real time, compares them with the digital twin model, calculates the cosine similarity, and compares it with a set threshold.

[0110] In another embodiment, considering the current realities faced by washing machine manufacturers—limited embedded hardware resources, long model training cycles, and high data privacy protection requirements—an alternative implementation plan that is easier to implement and has lower deployment costs is proposed. This involves using a lightweight digital twin model, which is not constructed using complex neural network models but rather based on a sliding window statistical method to build a "simplified digital twin model." Specifically, when using this simplified model, the mean and standard deviation of each feature are calculated from historical normal operation data as "benchmark features." Then, after real-time feature acquisition, the percentage deviation between the feature and the benchmark feature is calculated. If the deviation of a feature exceeds a preset dynamic threshold (e.g., ±3σ), an initial anomaly screening is triggered.

[0111] For the "simplified digital twin model", its anomaly judgment threshold can also be dynamically adjusted. Specifically, different baseline feature libraries are set according to the washing machine's operating stages (such as water filling, washing, and spin drying). Each stage corresponds to a set of independent benchmark features and thresholds. Then, the benchmark feature library is automatically updated every once in a while (such as weekly) to adapt to the aging trend of the equipment. The model is updated by the moving average method to avoid large fluctuations affecting the stability of the model.

[0112] Using the method provided in this embodiment, all core algorithms are deployed on the existing washing machine control motherboard (MCU + WiFi chip); no additional dedicated AI chip is required, saving BOM costs.

[0113] The above examples illustrate the method embodiments according to this application. The present invention also provides an anomaly detection device for a fabric processing equipment. Figure 5 This is a schematic diagram of the structure of an anomaly detection device in a fabric processing apparatus according to an embodiment of the present invention. (Refer to...) Figure 5 The abnormality detection device 700 of the fabric processing equipment includes the following modules.

[0114] Acquisition unit 701 acquires sound characteristic data and vibration characteristic data of the fabric processing equipment in real time during operation;

[0115] The determination unit 702 is used to compare the sound feature data with the vibration feature data and the normal feature distribution in the digital twin model to obtain the difference measurement index.

[0116] Processing unit 703 is used to determine whether there is an anomaly in the fabric processing equipment based on the difference measurement index and the adjusted anomaly judgment threshold.

[0117] The above describes the device embodiments of this application. For detailed descriptions of data, terms, nouns, specific execution processes of steps, technical problems and effects, alternative methods and combinations, please refer to the description in the method embodiments, which will not be repeated here.

[0118] This application also provides a computer program product including computer program instructions that, when executed by a processor, cause the processor to perform the steps in the anomaly detection method for a fabric processing apparatus according to various embodiments of this specification as described in the "Exemplary Methods" section above.

[0119] The computer program product can be written in any combination of one or more programming languages ​​to perform the operations of the embodiments of this specification. The programming languages ​​include object-oriented programming languages ​​such as Java and C++, as well as conventional procedural programming languages ​​such as the "C" language or similar programming languages.

[0120] This application also provides a computer-readable storage medium having a computer program stored thereon, the computer program being executed by a processor of the steps in the anomaly detection method of the fabric processing apparatus according to various embodiments of this specification as described in the "Exemplary Methods" section above.

[0121] This application also provides an electronic device, including a memory and a processor. The memory stores an anomaly detection method for a fabric processing device, and the processor is used to employ the anomaly detection method for the fabric processing device as described above when executing the anomaly detection method for the fabric processing device.

[0122] Specifically, such as Figure 6 As shown, the electronic device includes a processor 100, at least one communication bus 200, a user interface 300, at least one external communication interface 400, and a memory 500. The communication bus 200 is configured to enable communication between these components. The user interface 300 may include a display screen, and the external communication interface 400 may include standard wired and wireless interfaces. The memory 500 stores anomaly detection methods for fabric processing equipment. The processor 100 is used to employ the aforementioned methods when executing the anomaly detection methods for fabric processing equipment stored in the memory 500.

[0123] The descriptions of the above computer program products, computer-readable storage media, and electronic devices are similar to those of the above method embodiments, and have similar beneficial effects. For any technical details not disclosed in the computer program products, computer-readable storage media, and electronic devices of this application, please refer to the descriptions of the method embodiments of this application for understanding.

[0124] The sequence numbers or order of description of the embodiments in this application are for descriptive purposes only and do not represent the superiority or inferiority of the embodiments.

[0125] In the several embodiments provided in this application, it should be understood that the disclosed technical content can be implemented in other ways. The device embodiments described above are merely illustrative; for example, the division of units can be a logical functional division, and in actual implementation, there may be other division methods. For instance, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the displayed or discussed mutual coupling, direct coupling, or communication connection may be through some interfaces; the indirect coupling or communication connection between units or modules may be electrical or other forms.

[0126] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0127] Furthermore, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0128] In the above embodiments, implementation can be achieved, in whole or in part, through software, hardware, firmware, or any combination thereof. When implemented in software, it can be implemented, in whole or in part, as a computer program product. The computer program product includes one or more computer instructions. When the computer instructions are loaded and executed on a computer, all or part of the processes or functions described in the embodiments of this application are generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, the computer instructions can be transmitted from one website, computer, server, or data center to another via wired (e.g., coaxial cable, fiber optic, digital subscriber line (DSL)) or wireless (e.g., infrared, wireless, microwave, etc.) means. The computer-readable storage medium can be any available medium accessible to a computer, or a data storage device such as a server or data center that integrates one or more available media. The available medium can be a magnetic medium (e.g., floppy disk, hard disk, magnetic tape), an optical medium (e.g., digital versatile disc (DVD)), or a semiconductor medium (e.g., solid state disk (SSD)). It is worth noting that the computer-readable storage medium mentioned in the embodiments of this application can be a non-volatile storage medium; in other words, it can be a non-transient storage medium.

[0129] It should be noted that the information (including but not limited to user device information, user personal information, etc.), data (including but not limited to data used for analysis, stored data, displayed data, etc.), and signals involved in the embodiments of this application are all authorized by the user or fully authorized by all parties, and the collection, use, and processing of related data must comply with the relevant laws, regulations, and standards of the relevant countries and regions. For example, the scene data of the current frame in the 3D virtual scene involved in the embodiments of this application, the client's device information, and the scene interaction information are all obtained with full authorization.

[0130] The above description is only a preferred embodiment of this application. It should be noted that for those skilled in the art, several improvements and modifications can be made without departing from the principle of this application, and these improvements and modifications should also be considered within the scope of protection of this application.

Claims

1. An abnormality detection method of a fabric treatment apparatus, characterized by, The method comprises: Real-time acquisition of sound feature data and vibration feature data of the fabric treatment equipment during operation; Comparing the sound feature data and the vibration feature data with the corresponding normal feature distribution in the digital twin model to obtain a difference measurement index, wherein the normal feature distribution refers to a feature distribution obtained by training the digital twin model based on sound feature data and vibration feature data collected under a fault-free operation state of the fabric treatment equipment; Dynamically adjusting a pre-set abnormality judgment threshold in combination with the current working condition of the fabric treatment equipment, wherein the current working condition includes the aging condition and the use environment condition of the fabric treatment equipment; Determining whether the fabric treatment equipment is abnormal according to the difference measurement index and the adjusted abnormality judgment threshold; The comparison of the sound feature data and the vibration feature data with the corresponding normal feature distribution in the digital twin model to obtain a difference measurement index comprises: Calculating the Euclidean distance and / or cosine similarity between the sound feature data, the vibration feature data, and the corresponding normal feature distribution in the digital twin model; Determining the difference measurement index according to the calculation result; The determination of whether the fabric treatment equipment is abnormal according to the difference measurement index and the adjusted abnormality judgment threshold comprises: If the difference measurement index exceeds the abnormality judgment threshold, it is determined that the fabric treatment equipment is in an abnormal state.

2. The fabric treatment apparatus anomaly detection method according to claim 1, characterized in that, The device aging condition includes device operating load and device historical working condition data, and the use environment condition includes environmental temperature, environmental humidity, and environmental noise condition. The dynamic adjustment of the pre-set abnormality judgment threshold in combination with the current working condition of the fabric treatment equipment comprises: Inputting the device operating load and the device historical working condition data into a long short-term memory network to determine the aging degree of the fabric treatment equipment; Inputting the environmental temperature, the environmental humidity, and the environmental noise condition into an environmental interference convolutional neural network to determine the interference factors of the fabric treatment equipment; Dynamically adjusting the abnormality judgment threshold according to the aging degree and the interference factors.

3. The fabric treatment apparatus anomaly detection method according to claim 1, characterized in that, The real-time acquisition of sound feature data and vibration feature data of the fabric treatment equipment during operation comprises: Real-time acquisition of sound signals and vibration signals during operation through high-sensitivity acoustic sensors and vibration sensors installed inside the washing machine; Pretreatment of the sound signals and the vibration signals, and extraction of multi-dimensional features to obtain the sound feature data and the vibration feature data.

4. The fabric treatment apparatus anomaly detection method according to claim 1, characterized in that, The method further comprises: During the operation of the fabric treatment equipment, dynamically updating the parameters of the digital twin model through a sliding window mechanism to ensure that the model is consistent with the actual equipment state.

5. The fabric treatment apparatus anomaly detection method according to claim 1, characterized in that, The training method of the digital twin model comprises: Using normal operation data of the fabric treatment equipment under multiple working conditions as a training set, and extracting Mel frequency cepstrum coefficient features and vibration time domain statistical features of each frame of signal to construct an initial feature space; Using a K-means clustering algorithm to cluster the feature space and extract feature centers representing normal operation states; The feature centers are trained by using a support vector machine classifier to distinguish normal conditions from abnormal conditions, and a normal feature distribution is obtained.

6. The fabric treatment apparatus anomaly detection method according to claim 4, characterized in that, The digital twin model is deployed on a cloud server, and the method further comprises: Multi-device collaborative training and model sharing are realized through the cloud server; The performance of the digital twin model is analyzed and optimized through the cloud server, and the optimized model parameters are periodically obtained from the cloud server and deployed to the fabric treatment device.

7. The fabric treatment apparatus anomaly detection method according to claim 1, characterized in that, The digital twin model is a simplified digital twin model constructed based on a sliding window statistical method, and the corresponding normal feature distribution in the simplified digital twin model is a reference feature. The reference feature is used to calculate a deviation percentage with the sound feature data and the vibration feature data, and the deviation percentage is taken as the difference measurement index.

8. The fabric treatment apparatus anomaly detection method according to claim 2, characterized in that, The abnormality judgment threshold is dynamically adjusted according to the aging degree and the interference factor, including: The abnormality judgment threshold is adjusted according to the aging degree and the interference factor by the following formula: W = W1X1 + W2X2 + W3X3; Wherein, X1 is the aging degree, X2 is the interference factor, X3 is the abnormality judgment threshold, W is the adjusted abnormality judgment threshold, W1, W2, W3 are weight parameters of the aging degree, the interference factor and the abnormality judgment threshold respectively, and the sum of W1, W2, W3 is 1.

9. An electronic device, comprising: Including: A memory and a processor, which are communicatively connected between each other, the memory stores computer instructions, and the processor executes the computer instructions to perform the abnormality detection method of the fabric treatment device in any one of claims 1 to 8.

10. A fabric treatment apparatus characterised in that, It adopts the abnormality detection method of the fabric treatment device in any one of claims 1 to 8, or has the electronic device in claim 9.

Images