Anomaly detection systems, methods, devices, electronic equipment and media for smart homes

Abstract

This invention provides an anomaly monitoring system, method, device, electronic device, and medium for smart homes, belonging to the field of smart home technology. In the anomaly monitoring system, monitoring devices collect physical data from target devices and send it to a gateway device; generate status information based on fused data and send it to the gateway device; the gateway device performs fusion processing on physical data from multiple monitoring devices to obtain fused data; sends the fused data to multiple monitoring devices; and makes a comprehensive decision based on the received status information from multiple monitoring devices to obtain an anomaly judgment result. This invention improves the reliability of collected data by fusing physical data from multiple monitoring devices, reduces dependence on the gateway device and lowers processing latency by having the monitoring devices determine the status of the target device, and improves the accuracy of anomaly judgment results by using the status information determined by multiple monitoring devices for global judgment.

Description

Technical Field

[0001] This invention relates to the field of home appliance technology, and in particular to an anomaly monitoring system, method, device, electronic device, and readable storage medium for smart homes. Background Technology

[0002] Smart home systems connect various devices in the home through the Internet of Things (IoT) technology, enabling centralized control and intelligent management of appliances, significantly improving user convenience and home security. However, with the increasing variety and age of home appliances, potential malfunctions may arise. If these malfunctions are not detected and addressed promptly, they can not only affect the user experience but also pose safety hazards. Therefore, how to efficiently and accurately monitor the operating status of home devices has become an important research direction in the field of smart homes.

[0003] In related technologies, smart home systems typically connect various home devices through a home host, which can then acquire various status data of the home devices and analyze this data to determine if there are any abnormalities in the home devices, thereby ensuring the normal operation of the smart home system.

[0004] However, the above-mentioned anomaly detection methods rely too heavily on home computers, resulting in problems such as high latency and low reliability. Summary of the Invention

[0005] In view of the above problems, embodiments of the present invention are proposed to provide an anomaly monitoring system, method, apparatus, electronic device and readable storage medium for smart homes that overcomes or at least partially solves the above problems.

[0006] In a first aspect, embodiments of the present invention provide an anomaly monitoring system for smart homes, the anomaly monitoring system comprising multiple monitoring devices, target devices, and gateway devices; The monitoring device is used to collect physical data of the target device and send it to the gateway device; receive fused data sent by the gateway device, generate status information based on the fused data and send it to the gateway device; The gateway device is configured to perform fusion processing on physical data from multiple monitoring devices to obtain fused data; send the fused data to multiple monitoring devices; and make a comprehensive decision based on the received status information from multiple monitoring devices to obtain an anomaly judgment result.

[0007] Optionally, the gateway device is further configured to assign monitoring tasks to the monitoring device and send them to the monitoring device; the monitoring task is used to instruct the monitoring device to collect data from the corresponding target device; The monitoring device is also used to receive the monitoring task, determine the corresponding target device according to the monitoring task, and collect the physical data of the target device.

[0008] Optionally, the gateway device is specifically used to acquire historical acquisition quality information of the monitoring device and location information between the monitoring device and the target device; and to determine the monitoring task of the monitoring device based on the historical acquisition quality information and the location information.

[0009] Optionally, the physical data includes acoustic signature data and / or vibration data; the monitoring device is equipped with an acoustic signature acquisition device and / or a vibration acquisition device. The monitoring equipment is specifically used to collect the acoustic data of the target device through the acoustic data acquisition device; and / or to collect the vibration data of the target device through the vibration acquisition device.

[0010] Optionally, the gateway device is specifically configured to determine the weight of the corresponding monitoring device based on the historical acquisition quality information and the location information; and to perform weighted processing on the physical data collected by the monitoring device based on the weight of the monitoring device to obtain the fused data.

[0011] Optionally, a preset first diagnostic model is deployed on the monitoring device; The monitoring device is specifically used to extract features from the fused data to obtain feature information; The feature information is input into the first diagnostic model to obtain the status information output by the first diagnostic model.

[0012] Optionally, a preset second diagnostic model is deployed on the gateway device; The gateway device is further configured to acquire the feature information when the abnormal diagnosis conditions are met; input the feature information into the second diagnostic model to obtain the abnormal type and solution output by the second diagnostic model.

[0013] Optionally, the monitoring device is further configured to train the first diagnostic model using the physical data to obtain model parameter update information; encrypt the model parameter update information and send it to the gateway device; and update the first diagnostic model according to the model update knowledge sent by the gateway device. The gateway device is further configured to update the second diagnostic model based on encrypted model parameter update information from multiple monitoring devices, obtain model update knowledge, and send it to the monitoring devices.

[0014] Secondly, embodiments of the present invention provide an anomaly monitoring method for smart homes, applied to monitoring devices, the method comprising: The physical data of the target device is collected and sent to the gateway device. The physical data is used by the gateway device to generate fused data. The system receives fused data sent by the gateway device, generates status information based on the fused data, and sends it to the gateway device. The status information is used by the gateway device to make comprehensive decisions to obtain anomaly judgment results.

[0015] Optionally, the method further includes: The monitoring device receives a monitoring task, determines the corresponding target device based on the monitoring task, and collects physical data of the target device. The monitoring task is assigned by the gateway device and is used to instruct the monitoring device to collect data from the corresponding target device.

[0016] Optionally, the physical data includes acoustic signature data and / or vibration data; the monitoring device is equipped with an acoustic signature acquisition device and / or a vibration acquisition device. The physical data of the target device being collected includes: The voiceprint data of the target device is collected using the voiceprint acquisition device; and / or Vibration data of the target equipment are collected using the vibration acquisition device.

[0017] Optionally, a preset first diagnostic model is deployed on the monitoring device; The step of generating status information based on the fusion data sent by the gateway device includes: Feature extraction is performed on the fused data to obtain feature information; The feature information is input into the first diagnostic model to obtain the status information output by the first diagnostic model.

[0018] Optionally, a preset second diagnostic model is deployed on the gateway device; The method further includes: The first diagnostic model is trained using the physical data to obtain model parameter update information; The model parameter update information is encrypted and sent to the gateway device; the encrypted model parameter update information is used by the gateway device to update the second diagnostic model to obtain model update knowledge; The first diagnostic model is updated based on the model update knowledge sent by the gateway device.

[0019] Thirdly, embodiments of the present invention provide a method for detecting anomalies in smart homes, applied to a gateway device, the method comprising: Physical data from multiple monitoring devices are fused to obtain fused data; the physical data is collected by the monitoring devices from the target device. The fused data is sent to multiple monitoring devices; the fused data is used by the monitoring devices to generate status information. An anomaly judgment result is obtained by comprehensively making decisions based on the status information received from multiple monitoring devices.

[0020] Optionally, the method further includes: A monitoring task is assigned to the monitoring device and sent to the monitoring device; the monitoring task is used to instruct the monitoring device to collect data from the corresponding target device.

[0021] Optionally, assigning monitoring tasks to the monitoring equipment includes: Acquire historical data acquisition quality information of the monitoring device and location information between the monitoring device and the target device; The monitoring task of the monitoring equipment is determined based on the historical quality information and the location information.

[0022] Optionally, the step of fusing physical data from multiple monitoring devices to obtain the fused data includes: The weight of the corresponding monitoring device is determined based on the historical data collection quality information and the location information. The physical data collected by the monitoring devices are weighted according to their respective weights to obtain the fused data.

[0023] Optionally, a preset second diagnostic model is deployed on the gateway device; The method further includes: When the abnormal diagnostic conditions are met, feature information is acquired; the feature information is obtained by the monitoring device extracting features from the fused data. The feature information is input into the second diagnostic model to obtain the abnormality type and solution output by the second diagnostic model.

[0024] Optionally, a preset first diagnostic model is deployed on the monitoring device; The method further includes: The second diagnostic model is updated based on encrypted model parameter update information from multiple monitoring devices to obtain model update knowledge, which is then sent to the monitoring devices. The model parameter update information is obtained by the monitoring device through training the first diagnostic model using the physical data; the model update knowledge is used by the monitoring device to update the first diagnostic model.

[0025] Fourthly, embodiments of the present invention provide an anomaly monitoring device for smart homes, applied to monitoring equipment, the device comprising: The physical data acquisition module is used to acquire physical data of the target device and send it to the gateway device. The physical data is used by the gateway device to generate fused data. The status information generation module is used to receive the fused data sent by the gateway device, generate status information based on the fused data, and send it to the gateway device; the status information is used by the gateway device for comprehensive processing to obtain an anomaly judgment result.

[0026] Optionally, the device further includes the following modules: The target device determination module is used to receive a monitoring task, determine the corresponding target device according to the monitoring task, and collect physical data of the target device; the monitoring task is assigned by the gateway device and is used to instruct the monitoring device to collect data from the corresponding target device.

[0027] Optionally, the physical data includes acoustic signature data and / or vibration data; the monitoring device is equipped with an acoustic signature acquisition device and / or a vibration acquisition device. The physical data acquisition module includes the following sub-modules: The voiceprint data acquisition submodule is used to acquire voiceprint data of the target device through the voiceprint acquisition device; and / or The vibration data acquisition submodule is used to acquire vibration data of the target equipment through the vibration acquisition device.

[0028] Optionally, a preset first diagnostic model is deployed on the monitoring device; The status information generation module includes the following sub-modules: The feature extraction submodule is used to extract features from the fused data to obtain feature information; The status information output submodule is used to input the feature information into the first diagnostic model to obtain the status information output by the first diagnostic model.

[0029] Optionally, a preset second diagnostic model is deployed on the gateway device; The device also includes the following modules: The model parameter update information acquisition module is used to train the first diagnostic model using the physical data to obtain model parameter update information; The model parameter update information sending module is used to encrypt the model parameter update information and send it to the gateway device; the encrypted model parameter update information is used by the gateway device to update the second diagnostic model to obtain model update knowledge; The first diagnostic model update module is used to update the first diagnostic model based on the model update knowledge sent by the gateway device.

[0030] Fifthly, embodiments of the present invention provide an anomaly monitoring device for smart homes, applied to gateway devices, the device comprising: The fusion data generation module is used to fuse physical data from multiple monitoring devices to obtain fused data; the physical data is collected by the monitoring devices from the target device. A fused data transmission module is used to transmit the fused data to multiple monitoring devices; the fused data is used by the monitoring devices to generate status information. The anomaly judgment result determination module is used to make a comprehensive decision based on the received status information from multiple monitoring devices to obtain an anomaly judgment result.

[0031] Optionally, the device further includes the following modules: The monitoring task allocation module is used to allocate monitoring tasks to the monitoring device and send them to the monitoring device; the monitoring task is used to instruct the monitoring device to collect data from the corresponding target device.

[0032] Optionally, the monitoring task allocation module includes the following sub-modules: The information acquisition submodule is used to acquire historical data acquisition quality information of the monitoring device and location information between the monitoring device and the target device. The monitoring task determination submodule is used to determine the monitoring task of the monitoring device based on the historical acquisition quality information and the location information.

[0033] Optionally, the fused data generation module includes the following sub-modules: The weight determination submodule is used to determine the weight of the corresponding monitoring device based on the historical acquisition quality information and the location information; The weighted processing submodule is used to perform weighted processing on the physical data collected by the monitoring equipment according to the weight of the monitoring equipment to obtain the fused data.

[0034] Optionally, a preset second diagnostic model is deployed on the gateway device; The device also includes the following modules: The feature information acquisition module is used to acquire feature information when the abnormal diagnosis conditions are met; the feature information is obtained by the monitoring device through feature extraction of the fused data; The anomaly solution acquisition module is used to input the feature information into the second diagnostic model to obtain the anomaly type and solution output by the second diagnostic model.

[0035] Optionally, a preset first diagnostic model is deployed on the monitoring device; The device also includes the following modules: The model update knowledge acquisition module is used to update the second diagnostic model based on encrypted model parameter update information from multiple monitoring devices, obtain model update knowledge, and send it to the monitoring devices. The model parameter update information is obtained by the monitoring device through training the first diagnostic model using the physical data; the model update knowledge is used by the monitoring device to update the first diagnostic model.

[0036] In a sixth aspect, embodiments of the present invention provide an electronic device including a processor, a memory, and a program or instructions stored in the memory and executable on the processor, wherein the program or instructions, when executed by the processor, implement the steps of the anomaly monitoring method for smart homes as described in the second or third aspect.

[0037] In a seventh aspect, embodiments of the present invention provide a readable storage medium on which a program or instructions are stored, which, when executed by a processor, implement the steps of the anomaly monitoring method for smart homes as described in the second or third aspect.

[0038] The embodiments of the present invention have the following advantages: The anomaly monitoring system in this embodiment of the invention includes multiple monitoring devices, target devices, and gateway devices.

[0039] First, the monitoring devices collect physical data from the target device and send it to the gateway device. The gateway device then fuses the physical data from multiple monitoring devices to obtain fused data, which is then sent back to the multiple monitoring devices. By fusing physical data from multiple monitoring devices, the reliability of the collected data can be improved.

[0040] Then, the monitoring device generates status information based on the fused data sent by the gateway device and sends it to the gateway device. The target device status determined based on the fused information is more accurate, and having the monitoring device determine the target device status reduces reliance on the gateway device and lowers processing latency.

[0041] Finally, the gateway device makes a comprehensive decision based on the status information received from multiple monitoring devices to obtain an anomaly judgment result. By integrating the status information determined by multiple monitoring devices for the target device for global judgment, a more accurate anomaly judgment result can be determined, thereby achieving highly reliable and low-latency anomaly monitoring. Attached Figure Description

[0042] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the description of the embodiments of the present invention will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0043] Figure 1 This is a schematic diagram of the structure of an anomaly monitoring system for smart homes provided in an embodiment of the present invention; Figure 2 This is a flowchart illustrating the steps of an anomaly detection method for smart homes provided in an embodiment of the present invention; Figure 3 This is a flowchart of another method for detecting anomalies in smart homes provided by an embodiment of the present invention; Figure 4 This is a flowchart of another method for detecting anomalies in smart homes provided by an embodiment of the present invention; Figure 5 This is a structural block diagram of an anomaly monitoring device for smart homes provided in an embodiment of the present invention; Figure 6 This is a structural block diagram of another smart home anomaly monitoring device provided in an embodiment of the present invention. Detailed Implementation

[0044] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0045] The terms "first," "second," etc., used in the specification and claims of this invention are used to distinguish similar objects and are not used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that embodiments of the invention can be implemented in orders other than those illustrated or described herein. Furthermore, in the specification and claims, "and / or" indicates at least one of the connected objects, and the character " / " generally indicates that the preceding and following objects are in an "or" relationship.

[0046] Smart home systems connect various devices in the home through the Internet of Things (IoT) technology, enabling centralized control and intelligent management of appliances, significantly improving user convenience and home security. However, with the increasing variety and age of home appliances, potential malfunctions may arise. If these malfunctions are not detected and addressed promptly, they can not only affect the user experience but also pose safety hazards. Therefore, how to efficiently and accurately monitor the operating status of home devices has become an important research direction in the field of smart homes.

[0047] In related technologies, smart home systems typically connect various home devices through a home hub. This hub then acquires various status data from the devices and analyzes this data to determine if any abnormalities exist, ensuring the normal operation of the smart home system. However, this method of anomaly detection relies heavily on the home hub, resulting in issues such as high latency and low reliability.

[0048] One of the core concepts of this invention is to improve the reliability of collected data by integrating physical data from multiple monitoring devices. At the same time, the status of the target device is determined by the monitoring devices, reducing dependence on the gateway device and reducing processing latency. Then, global judgment is made by using the status information determined by multiple monitoring devices, thereby improving the accuracy of anomaly judgment results.

[0049] Figure 1 This is a schematic diagram of the structure of an anomaly monitoring system for smart homes provided in an embodiment of the present invention.

[0050] like Figure 1 As shown, the anomaly monitoring system includes multiple monitoring devices, target devices, and gateway devices.

[0051] The monitoring equipment is used to collect physical data of the target device and send it to the gateway device; it receives fused data sent by the gateway device, generates status information based on the fused data, and sends it to the gateway device. The gateway device is used to fuse physical data from multiple monitoring devices to obtain fused data; send the fused data to multiple monitoring devices; and make comprehensive decisions based on the status information received from multiple monitoring devices to obtain anomaly judgment results.

[0052] In real-world home scenarios, smart devices with edge computing capabilities, such as smart speakers, smart refrigerators, and smart air conditioners, can be used as monitoring devices to collect and analyze data. Smart routers and home servers can act as gateway devices to fuse data and make comprehensive decisions. Target devices refer to home appliances whose health status needs to be monitored; these can be either smart or non-smart devices. This anomaly monitoring system can fully utilize existing smart devices in the home to form a monitoring and sensing network covering the entire house, achieving health monitoring of both smart and non-smart devices at a relatively low cost.

[0053] Figure 2 This is a flowchart of the steps of an anomaly monitoring method for smart homes provided in an embodiment of the present invention.

[0054] In some embodiments, the gateway device is further configured to assign monitoring tasks to the monitoring device and send them to the monitoring device; the monitoring task is used to instruct the monitoring device to collect data from the corresponding target device. The monitoring equipment is also used to receive monitoring tasks, determine the corresponding target equipment based on the monitoring tasks, and collect physical data of the target equipment.

[0055] In the initial phase, the gateway device assigns monitoring tasks to the monitoring devices according to a preset task allocation strategy. This strategy comprehensively considers multiple factors and dynamically selects a monitoring combination including multiple monitoring devices for the target device, ensuring the accuracy of data collection while avoiding redundant collection or mutual interference. The monitoring tasks are then assigned to the corresponding monitoring devices, instructing them to perform physical data collection on the corresponding target devices.

[0056] In some embodiments, the gateway device is specifically used to acquire historical acquisition quality information of the monitoring device and location information between the monitoring device and the target device; and to determine the monitoring task of the monitoring device based on the historical acquisition quality information and the location information.

[0057] Specifically, the above task allocation strategy can be to allocate the monitoring task of the target device to a suitable monitoring device based on the historical data collection quality information of the monitoring device and the location information between the monitoring device and the target device.

[0058] The historical acquisition quality information includes historical data acquisition quality scores. The system will continuously record the historical data acquisition quality scores of each monitoring device. The score combines signal strength, clarity, and the proportion of valid data packets after cross-validation using the 3σ criterion. If a monitoring device submits low-quality or invalid data multiple times in the monitoring tasks of a specific target device, its acquisition quality score for that target device will be reduced.

[0059] Gateway devices can acquire the physical or network topology location information of all monitoring devices and target devices, thereby determining the location information between the monitoring devices and the target devices. It's understandable that physical data collected by devices that are closer to the target devices is generally less affected by environmental noise and has higher data quality.

[0060] When assigning monitoring tasks, the gateway device calculates the compatibility score between each target device and all surrounding monitoring devices. Specifically, this score can be a weighted sum of a distance factor and a data acquisition quality factor. The gateway device selects the monitoring device with the highest compatibility score for each target device as the primary acquisition node, responsible for core data acquisition; and selects 1-2 monitoring devices with the next highest scores as auxiliary and / or verification nodes. When the network topology changes, such as when the historical quality of a device or a node changes significantly, the monitoring tasks are automatically recalculated and reassigned.

[0061] In addition, the gateway device will adopt differentiated monitoring and data collection strategies for different target devices. For continuously running target devices such as refrigerators, periodic data collection will be implemented, such as collecting data every 10 minutes. For intermittently running target devices such as washing machines, event-triggered data collection will be used, such as initiating data collection based on energy consumption surge detection.

[0062] The allocation of monitoring tasks enables collaborative processing and task division among multiple monitoring devices. Centralized scheduling via the gateway device avoids data acquisition conflicts and redundancy between nodes, improving the overall efficiency of the system. Monitoring devices execute data acquisition according to task instructions, reducing real-time dependence on the gateway device and ensuring data acquisition quality.

[0063] In some embodiments, physical data includes acoustic fingerprint data and / or vibration data; the monitoring device is equipped with an acoustic fingerprint acquisition device and / or a vibration acquisition device. Monitoring equipment, specifically used to collect acoustic data of the target device through an acoustic data acquisition device; and / or to collect vibration data of the target device through a vibration acquisition device.

[0064] After receiving the monitoring task from the gateway device, the monitoring device identifies the target device to be monitored and activates the corresponding acoustic signature acquisition device and / or vibration acquisition device according to the task requirements to collect the physical data of the target device. The collected physical data is then sent to the gateway device for fusion processing.

[0065] Voiceprint data refers to the voiceprint signals generated by a target device. Different types of devices exhibit different acoustic characteristics under normal and abnormal conditions. For example, a refrigerator compressor emits a steady, low-frequency humming sound during normal operation, but may produce periodic abnormal noises when bearings wear out; a washing machine may produce abnormal vibration noise during the spin-drying stage. By collecting and analyzing voiceprint data, potential anomalies in the device can be effectively identified.

[0066] Vibration data refers to the vibration signals generated by the target equipment. When the mechanical components of the equipment operate, they produce vibrations with specific frequencies and amplitudes. When abnormalities occur, the time-domain waveform, frequency-domain distribution, and energy characteristics of the vibration signal will change. For example, wear on the fan bearings of an air conditioner outdoor unit can lead to increased vibration amplitude and the appearance of high-frequency harmonics; aging of the vibration damping system in a washing machine's spin-dry tub can cause the vibration waveform to exhibit irregular impact characteristics. Vibration data and acoustic signature data complement each other, reflecting the equipment's condition from different physical dimensions and providing more comprehensive data for subsequent diagnostics.

[0067] In some embodiments, the physical data also includes energy consumption data, which can be collected by connecting to the target device via a smart socket.

[0068] In some embodiments, the gateway device is specifically used to determine the weight of the corresponding monitoring device based on historically collected quality information and location information; and to perform weighted processing on the physical data collected by the monitoring device according to the weight of the monitoring device to obtain fused data.

[0069] The gateway device receives physical data collected from different monitoring devices on the same target device within the same time period. Due to differences in the relative positions and sensor sensitivities between the monitoring devices and the target device, the quality of the collected data varies. Therefore, this embodiment introduces a weighted fusion mechanism, assigning a weight W to each monitoring device node. i The weight calculation formula is as follows:

[0070] Among them, D i Q is a comprehensive distance index for node i. A larger value indicates a greater distance or a weaker signal. i It represents the recent historical acquisition accuracy of node i, ranging from 0 to 1. α and β are adjustable weighting coefficients that satisfy α+β=1, used to balance the importance of distance and quality in this fusion.

[0071] After determining the weights of each monitoring device, the physical data from each device are weighted to obtain a weighted average fused signal. Specifically, the acoustic signature data and vibration data are weighted separately to obtain corresponding acoustic signature fused data and vibration fused data. A cross-validation mechanism based on the 3σ criterion can also be applied to automatically identify and remove abnormal data packets caused by factors such as instantaneous environmental noise, further ensuring data quality. The fused data can suppress noise caused by poor node location or instantaneous interference, resulting in a higher signal-to-noise ratio and reliability compared to the original data collected from a single node. This ensures that subsequent analysis can be based on high-quality data, improving the reliability of subsequent diagnostics.

[0072] After obtaining the fused data, the gateway device sends the fused data to the monitoring device. The monitoring device has edge computing capabilities and can make a preliminary judgment on the status of the target device based on the fused data.

[0073] In some embodiments, a preset first diagnostic model is deployed on the monitoring device; Monitoring equipment is specifically used to extract features from fused data to obtain feature information; The feature information is input into the first diagnostic model to obtain the status information output by the first diagnostic model.

[0074] After receiving the fused data from the gateway device, the monitoring equipment extracts features from the fused data to obtain feature information. For acoustic signature signals, features such as Mel-frequency cepstral coefficients and spectral centroids can be extracted. For vibration signals, vibration features such as time-domain peak values ​​and frequency-domain energy can be extracted. For energy consumption data, energy consumption features such as real-time power and fluctuation coefficients can be extracted.

[0075] The above feature information is input into the first diagnostic model. Typically, the first diagnostic model deployed on the monitoring device is a lightweight classification model, such as a Support Vector Machine (SVM), which is characterized by a small number of parameters, fast inference speed, and low computational resource consumption. Before deployment, the first diagnostic model is pre-trained using a labeled sample library containing various device states (normal state and multiple abnormal states) to enable it to have a preliminary ability to identify abnormal states. After the feature information is input into the first diagnostic model, its output state information is obtained. The state information includes a preliminary judgment result on whether the target device has an abnormal state, as well as the confidence level of this judgment result.

[0076] The aforementioned status information is securely transmitted to the gateway device via the local area network. The gateway device then makes a comprehensive decision based on the status information received from multiple monitoring devices to arrive at the final anomaly assessment result. Specifically, the gateway device employs a weighted aggregation algorithm, where the weights are dynamically adjusted based on the recent diagnostic accuracy of the monitoring devices. For example, nodes with high accuracy have a weight between 0.3 and 0.4. All preliminary assessment results are integrated to form a comprehensive decision, which is then immediately pushed to the user. The final anomaly assessment result categorizes the device health status into four levels: "Normal," "Level 1 Warning (Minor Anomaly)," "Level 2 Warning (Moderate Anomaly)," and "Level 3 Warning (Severe Anomaly)."

[0077] In some embodiments, a preset second diagnostic model is deployed on the gateway device; The gateway device is also used to acquire feature information when the abnormal diagnosis conditions are met; input the feature information into the second diagnostic model to obtain the abnormal type and solution output by the second diagnostic model.

[0078] To enhance the ability to identify complex faults, a preset second diagnostic model is deployed on the gateway device. When abnormal diagnostic conditions are met, such as when the preliminary diagnostic results output by the monitoring device indicate that the target device has an anomaly and the confidence level of the anomaly exceeds a preset threshold, or when the anomaly level obtained by the gateway device after comprehensive decision-making reaches a level 2 warning or above, or when it is actively triggered by the user, such as when requesting a deep inspection of a device through a mobile APP, the second diagnostic model can be activated to perform deep diagnosis.

[0079] As an example, the second diagnostic model can be a hybrid neural network model based on CNN-LSTM. The input to this model is a multi-dimensional feature matrix composed of acoustic signature feature vectors, vibration feature vectors, and energy consumption time-series data. The Convolutional Neural Network (CNN) layer is responsible for extracting local correlations of features, while the Long Short-Term Memory (LSTM) layer captures the temporal patterns of equipment state changes. The model output can accurately classify various anomaly types, such as bearing wear, refrigerant micro-leakage, and poor circuit contact. Before deployment, the model is pre-trained using a rich fault sample library and continuously optimized through a federated learning mechanism during subsequent operation to ensure its accurate identification of complex and latent faults. The second diagnostic model takes a three-dimensional tensor as input, with dimensions of [time step T, feature dimension F, number of channels C]. Here, T represents continuous data points within a time window; F includes various features extracted from acoustic signature, vibration, and energy consumption data; and C can be used to distinguish different types of data sources.

[0080] The hierarchical structure of the second diagnostic model includes an input layer, a CNN module, a feature remodeling layer, an LSTM module, an attention mechanism layer, and a fully connected output layer.

[0081] The input layer receives the aforementioned three-dimensional feature tensor. The CNN module is used for spatial feature extraction, employing multi-scale one-dimensional convolutional layers. Multiple convolutional kernels of different sizes are used in parallel to extract local features within different receptive fields, simultaneously capturing short-term impact (e.g., abnormal noise) and long-term modal (e.g., resonance) features. The feature reshaping layer converts the feature maps output by the CNN into a sequence format suitable for temporal network processing. The LSTM module is used for temporal feature extraction, consisting of 2-3 stacked LSTM layers, to learn the long-term temporal dependencies and evolution patterns of device state features. For the attention mechanism layer, an attention layer is added after the LSTM layers, enabling the model to automatically focus on the most relevant time steps and feature dimensions related to the fault, improving the ability to identify complex and latent faults. Finally, through fully connected layers and the Softmax function, the output outputs probability distributions corresponding to normal conditions and various specific anomaly types.

[0082] During the previous diagnostic process, each monitoring device extracted feature information from the fused data, including acoustic signature features, vibration features, and energy consumption features. This feature information can be cached locally on the monitoring devices or in the gateway device. When the aforementioned in-depth diagnostics are needed, the gateway device can request this feature information from each monitoring device or read it directly from its local cache. After the feature information is input into the second diagnostic model, the probability distribution of specific anomaly types and solutions output by the model are pushed to the user as supplementary and refined information for the high-level early warning, improving the operability of the early warning and its maintenance guidance value.

[0083] In some embodiments, the monitoring device is further configured to train the first diagnostic model using physical data to obtain model parameter update information; encrypt the model parameter update information and send it to the gateway device; and update the first diagnostic model according to the model update knowledge sent by the gateway device. The gateway device is also used to update the second diagnostic model based on encrypted model parameter update information from multiple monitoring devices, obtain model update knowledge, and send it to the monitoring devices.

[0084] Model parameter update information refers to the gradient of model parameters calculated by the backpropagation algorithm after the monitoring device incrementally trains the first diagnostic model. This model parameter update information can specifically be a gradient vector or parameter adjustment amount, allowing the gateway device to optimize the model by integrating the local learning results of each monitoring device without accessing the original data.

[0085] Model update knowledge refers to lightweight information extracted by the gateway device from the second diagnostic model through knowledge distillation, which can be used to update the first diagnostic model. Specifically, this model update knowledge can be soft labels or feature maps, used by lightweight models with different structures, such as the first diagnostic model, to learn the decision-making capabilities of complex models, such as the second diagnostic model.

[0086] This embodiment introduces a localized federated learning mechanism, enabling co-evolution and knowledge transfer between the first diagnostic model of the monitoring equipment and the second diagnostic model of the gateway device. Without disclosing the user's original data, the system can continuously learn from real-world data and optimize its diagnostic capabilities. Specifically, the monitoring equipment can train and update its first diagnostic model based on local physical data and send updated model parameters to the gateway device. The gateway device, in turn, can use this updated model parameters to synthesize the local learning results of each monitoring device and globally update the second diagnostic model. Furthermore, the gateway device can generate updated model knowledge based on the updated second diagnostic model and send it to each monitoring device, allowing each monitoring device to further update its first diagnostic model based on this updated knowledge, achieving iterative optimization.

[0087] After obtaining user authorization, the monitoring device incrementally trains the locally deployed first diagnostic model using newly collected physical data. The physical data serves as training samples, and its corresponding labels can be: diagnostic results output by the gateway device or the actual device status reported by the user. In some embodiments, the monitoring device inputs the collected physical data into the first diagnostic model, calculates the loss value between the model output and the true label, and then calculates the gradient information through a backpropagation algorithm. The gradient information reflects the direction and magnitude of the model parameters that need to be adjusted and is the core content of the model parameter update information.

[0088] To protect user privacy and prevent model parameter update information from being stolen or its original characteristics from being deduced during transmission, the monitoring device encrypts the model parameter update information before sending it to the gateway device. As an example, the Paillier homomorphic encryption algorithm can be used for encryption. The monitoring device then sends the encrypted model parameter update information to the gateway device via the home LAN.

[0089] The gateway device receives encrypted model parameter update information from multiple monitoring devices, securely aggregates this information, and then updates the second diagnostic model using the aggregated gradient. This allows the second diagnostic model to securely achieve global optimization by integrating the local learning results of multiple monitoring devices without accessing the original physical data.

[0090] After the second diagnostic model is updated, the gateway device transmits the updated knowledge to each monitoring device. Since the first and second diagnostic models on the monitoring devices typically differ structurally—the first is a lightweight model, while the second is a complex model—their parameters cannot be directly synchronized. Therefore, a knowledge distillation mechanism is introduced. The gateway device uses the updated second diagnostic model as the teacher model, extracts its core knowledge through knowledge distillation, generates model update knowledge, and sends it to each monitoring device. Upon receiving the model update knowledge from the gateway device, the monitoring devices update their local first diagnostic models accordingly.

[0091] The anomaly monitoring system in this embodiment of the invention includes multiple monitoring devices, target devices, and gateway devices.

[0092] Monitoring devices collect physical data from target devices and send it to a gateway device. The gateway device then fuses the physical data from multiple monitoring devices to obtain fused data, which is then sent back to the other monitoring devices. Fusing physical data from multiple monitoring devices improves the reliability of the collected data. The monitoring devices generate status information based on the fused data sent by the gateway device and send it to the gateway device as well. Determining the target device's status based on the fused information is more accurate, and having the monitoring devices determine the target device's status reduces reliance on the gateway device and lowers processing latency. The gateway device makes a comprehensive decision based on the status information received from multiple monitoring devices to arrive at an anomaly detection result. By comprehensively analyzing the status information determined by multiple monitoring devices for the target device, a more accurate anomaly detection result can be determined, achieving highly reliable and low-latency anomaly monitoring.

[0093] Figure 3 This is a flowchart of another method for anomaly detection in smart homes provided by an embodiment of the present invention.

[0094] like Figure 3 As shown in the figure, this embodiment of the invention provides a method for anomaly detection in smart homes, applied to monitoring devices. The method may specifically include the following steps: Step 301: Collect physical data of the target device and send it to the gateway device. The physical data is used by the gateway device to generate fused data. Prior to step 301, this method further includes: The monitoring device receives monitoring tasks, determines the corresponding target devices based on the monitoring tasks, and collects physical data from the target devices. The monitoring tasks are assigned by the gateway device and are used to instruct the monitoring device to collect data from the corresponding target devices.

[0095] In the initial phase, the gateway device assigns monitoring tasks to the monitoring devices according to a preset task allocation strategy. This strategy comprehensively considers multiple factors and dynamically selects a monitoring combination including multiple monitoring devices for the target device, ensuring the accuracy of data collection. After receiving a monitoring task, the monitoring device determines the corresponding target device based on the task to collect its physical data. In some embodiments, physical data includes acoustic fingerprint data and / or vibration data; the monitoring device is equipped with an acoustic fingerprint acquisition device and / or a vibration acquisition device. The physical data of the target device being collected includes: Acquire acoustic data of the target device using an acoustic data acquisition device; and / or acquire vibration data of the target device using a vibration data acquisition device.

[0096] After receiving a monitoring task from the gateway device, the monitoring equipment identifies the target device to be monitored and activates the corresponding acoustic signature acquisition device and / or vibration acquisition device according to the task requirements to collect physical data of the target device. The collected physical data is then sent to the gateway device for fusion processing. Vibration data and acoustic signature data complement each other, reflecting the status of the device from different physical dimensions and providing more comprehensive data for subsequent diagnosis.

[0097] In some embodiments, the physical data also includes energy consumption data, which can be collected by connecting to the target device via a smart socket.

[0098] Step 302: Receive the fusion data sent by the gateway device, generate status information based on the fusion data, and send it to the gateway device; the status information is used by the gateway device to make comprehensive decisions to obtain anomaly judgment results.

[0099] In some embodiments, a preset first diagnostic model is deployed on the monitoring device; The step of generating status information based on the fusion data sent by the gateway device includes: Feature extraction is performed on the fused data to obtain feature information; the feature information is then input into the first diagnostic model to obtain the state information output by the first diagnostic model.

[0100] After receiving the fused data from the gateway device, the monitoring equipment extracts features from the fused data to obtain feature information. For acoustic signature signals, features such as Mel-frequency cepstral coefficients and spectral centroids can be extracted. For vibration signals, vibration features such as time-domain peak values ​​and frequency-domain energy can be extracted. For energy consumption data, energy consumption features such as real-time power and fluctuation coefficients can be extracted.

[0101] The above feature information is input into the first diagnostic model to obtain its output status information. The status information includes a preliminary judgment result on whether the target device has an abnormal state, and the confidence level of the judgment result. This status information is securely sent to the gateway device through the local area network. The gateway device makes a comprehensive decision based on the status information received from multiple monitoring devices to obtain the final anomaly judgment result.

[0102] In some embodiments, a preset second diagnostic model is deployed on the gateway device; The method further includes: The first diagnostic model is trained using physical data to obtain model parameter update information; the model parameter update information is encrypted and sent to the gateway device; the encrypted model parameter update information is used by the gateway device to update the second diagnostic model to obtain model update knowledge; the first diagnostic model is updated based on the model update knowledge sent by the gateway device.

[0103] This embodiment introduces a localized federated learning mechanism, which enables co-evolution and knowledge transfer between the first diagnostic model of the monitoring device and the second diagnostic model of the gateway device. Without disclosing the user's original data, the system can continuously learn from actual data and continuously optimize its diagnostic capabilities.

[0104] After obtaining user authorization, the monitoring device incrementally trains the locally deployed first diagnostic model using newly collected physical data to obtain updated model parameters. To protect user privacy, the updated model parameters are encrypted before being sent to the gateway device. Upon receiving the encrypted updated model parameters from multiple monitoring devices, the gateway device securely aggregates the information and then updates the second diagnostic model using the aggregated gradient. After the second diagnostic model is updated, the gateway device transmits the updated knowledge to each monitoring device, extracts its core knowledge through knowledge distillation, generates model update knowledge, and sends it to each monitoring device. Upon receiving the model update knowledge from the gateway device, the monitoring device updates its local first diagnostic model accordingly, thus achieving bidirectional optimization.

[0105] Figure 4 This is a flowchart of another method for anomaly detection in smart homes provided by an embodiment of the present invention.

[0106] like Figure 4 As shown in the figure, this embodiment of the invention provides a method for anomaly detection in smart homes, applied to gateway devices. The method may specifically include the following steps: Step 401: The physical data from multiple monitoring devices are fused to obtain fused data; the physical data is collected by the monitoring devices from the target device. In some embodiments, between steps 401, the method further includes: Assign monitoring tasks to the monitoring equipment and send them to the monitoring equipment; the monitoring tasks are used to instruct the monitoring equipment to collect data from the corresponding target equipment.

[0107] In the initial phase, the gateway device assigns monitoring tasks to the monitoring devices according to a preset task allocation strategy. This strategy comprehensively considers multiple factors and dynamically selects a monitoring combination including multiple monitoring devices for the target device, ensuring the accuracy of data collection while avoiding redundant collection or mutual interference. The monitoring tasks are then assigned to the corresponding monitoring devices, instructing them to perform physical data collection on the corresponding target devices.

[0108] In some embodiments, assigning monitoring tasks to the monitoring equipment includes: Acquire historical data acquisition quality information from the monitoring equipment and location information between the monitoring equipment and the target equipment; determine the monitoring tasks of the monitoring equipment based on the historical data acquisition quality information and location information.

[0109] Specifically, the task allocation strategy described above can be as follows: based on the historical acquisition quality information of the monitoring devices and the location information between the monitoring devices and the target devices, the monitoring tasks of the target devices are assigned to appropriate monitoring devices. The historical acquisition quality information includes historical data acquisition quality scores. The system continuously records the historical data acquisition quality scores for each monitoring device, which integrate signal strength, clarity, and the proportion of valid data packets after 3σ cross-validation. Additionally, the gateway device can obtain the physical or network topology location information of all monitoring devices and the target device, thereby determining the location information between the monitoring devices and the target device. It is understood that the physical data collected by devices that are closer to the target device is generally less affected by environmental noise and has higher data quality.

[0110] When assigning monitoring tasks, the gateway device calculates the compatibility score between each target device and all surrounding monitoring devices. It then selects the monitoring device with the highest compatibility score for each target device as the primary data acquisition node, responsible for core data collection; and selects 1-2 monitoring devices with the second-highest scores as auxiliary and / or verification nodes. When the network topology changes, such as when the historical quality of a device or node changes significantly, the monitoring tasks are automatically recalculated and reassigned.

[0111] The allocation of monitoring tasks enables collaborative processing and task division among multiple monitoring devices. Centralized scheduling via the gateway device avoids data acquisition conflicts and redundancy between nodes, improving the overall efficiency of the system. Monitoring devices execute data acquisition according to task instructions, reducing real-time dependence on the gateway device and ensuring data acquisition quality.

[0112] After receiving the monitoring task from the gateway device, the monitoring device determines the target device to be monitored and starts the corresponding acquisition device to collect the physical data of the target device according to the task requirements. The collected physical data is then sent to the gateway device for fusion processing.

[0113] In some embodiments, the fusion processing of physical data from multiple monitoring devices to obtain fused data includes: Based on historical quality and location information, the weights of the corresponding monitoring devices are determined; based on the weights of the monitoring devices, the physical data collected by the monitoring devices are weighted to obtain fused data.

[0114] The gateway device receives physical data collected from different monitoring devices on the same target device within the same time period. Due to differences in the relative positions and sensor sensitivities between the monitoring devices and the target device, the quality of the collected data varies. Therefore, this embodiment introduces a weighted fusion mechanism, assigning a weight W to each monitoring device node. i The weight calculation formula is as follows:

[0115] Among them, D i Q is a comprehensive distance index for node i. A larger value indicates a greater distance or a weaker signal. i It represents the recent historical acquisition accuracy of node i, ranging from 0 to 1. α and β are adjustable weighting coefficients that satisfy α+β=1, used to balance the importance of distance and quality in this fusion.

[0116] After determining the weights of each monitoring device, the physical data from each device are weighted to obtain a weighted average of the fused data. The fused data can suppress noise caused by poor node location or transient interference, resulting in a higher signal-to-noise ratio and higher reliability compared to the original data collected from a single node. This ensures that subsequent analysis can be based on high-quality data, improving the reliability of subsequent diagnostics.

[0117] Step 402: The fused data is sent to multiple monitoring devices; the fused data is used by the monitoring devices to generate status information. After receiving the fused data, the gateway device sends it to the monitoring device. The monitoring device, equipped with edge computing capabilities, can make a preliminary judgment on the status of the target device based on the fused data and generate status information. The relevant content of the status information has been explained in step 302 above and will not be repeated here.

[0118] Step 403: Make a comprehensive decision based on the status information received from multiple monitoring devices to obtain the anomaly judgment result.

[0119] Status information is securely transmitted to the gateway device via the local area network. The gateway device then makes a comprehensive decision based on the status information received from multiple monitoring devices to arrive at the final anomaly assessment result. Specifically, the gateway device employs a weighted aggregation algorithm, where the weights are dynamically adjusted based on the recent diagnostic accuracy of the monitoring devices. For example, nodes with high accuracy have a weight between 0.3 and 0.4. All preliminary assessment results are integrated to form a comprehensive decision, which is then immediately pushed to the user. The final anomaly assessment result categorizes the device health status into four levels: "Normal," "Level 1 Warning (Minor Anomaly)," "Level 2 Warning (Moderate Anomaly)," and "Level 3 Warning (Severe Anomaly)."

[0120] In some embodiments, a preset second diagnostic model is deployed on the gateway device; The method further includes: When the abnormal diagnosis conditions are met, feature information is acquired; the feature information is obtained by the monitoring equipment through feature extraction of the fused data; the feature information is input into the second diagnostic model to obtain the abnormal type and solution output by the second diagnostic model.

[0121] To enhance the ability to identify complex faults, a preset second diagnostic model is deployed on the gateway device. When abnormal diagnostic conditions are met, such as when the preliminary diagnostic results output by the monitoring device indicate that the target device has an anomaly and the confidence level of the anomaly exceeds a preset threshold, or when the anomaly level obtained by the gateway device after comprehensive decision-making reaches a level 2 warning or above, or when it is actively triggered by the user, such as when requesting a deep inspection of a device through a mobile APP, the second diagnostic model can be activated to perform deep diagnosis.

[0122] As an example, the second diagnostic model can be a hybrid neural network model based on CNN-LSTM. During the previous diagnostic process, each monitoring device has extracted feature information from the fused data, including acoustic signature features, vibration features, and energy consumption features. This feature information can be cached locally on the monitoring devices or in the gateway device. When the above-mentioned abnormal diagnostic conditions are met, the gateway device can request these feature information from each monitoring device or read them directly from its local cache. After inputting the feature information into the second diagnostic model, the probability distribution of the specific abnormal type and the solution output by the model serve as supplementary and refined information to the high-level early warning, and are pushed to the user to improve the operability and maintenance guidance value of the early warning. In some embodiments, a preset first diagnostic model is deployed on the monitoring device; The method further includes: The second diagnostic model is updated based on encrypted model parameter update information from multiple monitoring devices, and the model update knowledge is obtained and sent to the monitoring devices. The model parameter update information is obtained by the monitoring devices through training the first diagnostic model with physical data. The model update knowledge is used by the monitoring devices to update the first diagnostic model.

[0123] To achieve co-evolution and knowledge transfer between the first diagnostic model of the monitoring equipment and the second diagnostic model of the gateway equipment, and to enable the system to continuously learn from actual data and optimize its diagnostic capabilities without disclosing the user's original data, this embodiment introduces a localized federated learning mechanism.

[0124] Once user authorization is obtained, the monitoring device uses the newly collected physical data to incrementally train the locally deployed first diagnostic model and obtain updated model parameter information.

[0125] To protect user privacy and prevent the model parameter update information from being stolen or its original characteristics from being revealed during transmission, the monitoring device encrypts the model parameter update information before sending it to the gateway device, and then sends the encrypted model parameter update information to the gateway device through the home LAN.

[0126] The gateway device receives encrypted model parameter update information from multiple monitoring devices, securely aggregates the encrypted model parameter update information, and then updates the second diagnostic model through the aggregated gradient. This enables the second diagnostic model to integrate the local learning results of multiple monitoring devices without touching the original physical data, thus achieving secure global optimization.

[0127] After the second diagnostic model is updated, the gateway device transmits the updated knowledge to each monitoring device. Since the first and second diagnostic models on the monitoring devices typically differ structurally—the first is a lightweight model, while the second is a complex model—they cannot directly synchronize their parameters. Therefore, a knowledge distillation mechanism is introduced. The gateway device uses the updated second diagnostic model as the teacher model, extracts its core knowledge through knowledge distillation, generates model update knowledge, and sends it to each monitoring device. Upon receiving the model update knowledge from the gateway device, the monitoring devices update their local first diagnostic models accordingly, thus achieving bidirectional optimization.

[0128] It should be noted that, for the sake of simplicity, the method embodiments are all described as a series of actions. However, those skilled in the art should understand that the embodiments of the present invention are not limited to the described order of actions, because according to the embodiments of the present invention, some steps can be performed in other orders or simultaneously. Furthermore, those skilled in the art should also understand that the embodiments described in the specification are preferred embodiments, and the actions involved are not necessarily essential to the embodiments of the present invention.

[0129] Figure 5 This is a structural block diagram of an anomaly monitoring device for smart homes provided in an embodiment of the present invention.

[0130] like Figure 5 As shown in the figure, an anomaly monitoring device for smart homes provided in this embodiment of the invention is applied to monitoring equipment and may specifically include the following modules: The physical data acquisition module 501 is used to acquire physical data of the target device and send it to the gateway device. The physical data is used by the gateway device to generate fused data. The status information generation module 502 is used to receive the fused data sent by the gateway device, generate status information based on the fused data, and send it to the gateway device; the status information is used by the gateway device for comprehensive processing to obtain an anomaly judgment result.

[0131] In some embodiments, the device further includes the following modules: The target device determination module is used to receive a monitoring task, determine the corresponding target device according to the monitoring task, and collect physical data of the target device; the monitoring task is assigned by the gateway device and is used to instruct the monitoring device to collect data from the corresponding target device.

[0132] In some embodiments, the physical data includes acoustic signature data and / or vibration data; the monitoring device is equipped with an acoustic signature acquisition device and / or a vibration acquisition device; The physical data acquisition module includes the following sub-modules: The voiceprint data acquisition submodule is used to acquire voiceprint data of the target device through the voiceprint acquisition device; and / or The vibration data acquisition submodule is used to acquire vibration data of the target equipment through the vibration acquisition device.

[0133] In some embodiments, a preset first diagnostic model is deployed on the monitoring device; The status information generation module includes the following sub-modules: The feature extraction submodule is used to extract features from the fused data to obtain feature information; The status information output submodule is used to input the feature information into the first diagnostic model to obtain the status information output by the first diagnostic model.

[0134] In some embodiments, a preset second diagnostic model is deployed on the gateway device; The device also includes the following modules: The model parameter update information acquisition module is used to train the first diagnostic model using the physical data to obtain model parameter update information; The model parameter update information sending module is used to encrypt the model parameter update information and send it to the gateway device; the encrypted model parameter update information is used by the gateway device to update the second diagnostic model to obtain model update knowledge; The first diagnostic model update module is used to update the first diagnostic model based on the model update knowledge sent by the gateway device.

[0135] Figure 6 This is a structural block diagram of another smart home anomaly monitoring device provided in an embodiment of the present invention.

[0136] like Figure 6 As shown in the figure, an anomaly monitoring device for smart homes provided in this embodiment of the invention is applied to a gateway device and may specifically include the following modules: The fusion data generation module 601 is used to fuse physical data from multiple monitoring devices to obtain fused data; the physical data is collected by the monitoring devices from the target device. The fused data sending module 602 is used to send the fused data to multiple monitoring devices; the fused data is used by the monitoring devices to generate status information. The anomaly judgment result determination module 603 is used to make a comprehensive decision based on the received status information from multiple monitoring devices to obtain an anomaly judgment result.

[0137] In some embodiments, the device further includes the following modules: The monitoring task allocation module is used to allocate monitoring tasks to the monitoring device and send them to the monitoring device; the monitoring task is used to instruct the monitoring device to collect data from the corresponding target device.

[0138] In some embodiments, the monitoring task allocation module includes the following sub-modules: The information acquisition submodule is used to acquire historical data acquisition quality information of the monitoring device and location information between the monitoring device and the target device. The monitoring task determination submodule is used to determine the monitoring task of the monitoring device based on the historical acquisition quality information and the location information.

[0139] In some embodiments, the fused data generation module includes the following sub-modules: The weight determination submodule is used to determine the weight of the corresponding monitoring device based on the historical acquisition quality information and the location information; The weighted processing submodule is used to perform weighted processing on the physical data collected by the monitoring equipment according to the weight of the monitoring equipment to obtain the fused data.

[0140] In some embodiments, a preset second diagnostic model is deployed on the gateway device; The device also includes the following modules: The feature information acquisition module is used to acquire feature information when the abnormal diagnosis conditions are met; the feature information is obtained by the monitoring device through feature extraction of the fused data; The anomaly solution acquisition module is used to input the feature information into the second diagnostic model to obtain the anomaly type and solution output by the second diagnostic model.

[0141] In some embodiments, a preset first diagnostic model is deployed on the monitoring device; The device also includes the following modules: The model update knowledge acquisition module is used to update the second diagnostic model based on encrypted model parameter update information from multiple monitoring devices, obtain model update knowledge, and send it to the monitoring devices. The model parameter update information is obtained by the monitoring device through training the first diagnostic model using the physical data; the model update knowledge is used by the monitoring device to update the first diagnostic model.

[0142] As the apparatus embodiment is basically similar to the method embodiment, it is described in a relatively simple manner. For relevant details, please refer to the description of the method embodiment.

[0143] This invention also provides an electronic device, including a processor, a memory, and a program or instructions stored in the memory and executable on the processor. When the program or instructions are executed by the processor, they implement the various processes of the above-described smart home anomaly monitoring method embodiments and achieve the same technical effects. To avoid repetition, they will not be described again here.

[0144] It should be noted that the electronic devices in the embodiments of the present invention include the mobile electronic devices and non-mobile electronic devices described above.

[0145] This invention also provides a readable storage medium storing a program or instructions. When the program or instructions are executed by a processor, they implement the various processes of the above-described smart home anomaly monitoring method embodiments and achieve the same technical effects. To avoid repetition, they will not be described again here.

[0146] The processor is the processor in the electronic device described in the above embodiments. The readable storage medium includes computer-readable storage media, such as computer read-only memory (ROM), random access memory (RAM), magnetic disk, or optical disk.

[0147] The various embodiments in this specification are described in a progressive manner, with each embodiment focusing on the differences from other embodiments. The same or similar parts between the various embodiments can be referred to each other.

[0148] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, apparatus, or computer program products. Therefore, embodiments of the present invention can take the form of entirely hardware embodiments, entirely software embodiments, or embodiments combining software and hardware aspects. Furthermore, embodiments of the present invention can take the form of computer program products implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0149] Embodiments of the present invention are described with reference to flowchart illustrations and / or block diagrams of methods, terminal devices (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, as well as combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing terminal device to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing terminal device, generate instructions for implementing the flowchart illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0150] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing terminal device to operate in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0151] These computer program instructions can also be loaded onto a computer or other programmable data processing terminal equipment, causing a series of operational steps to be performed on the computer or other programmable terminal equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable terminal equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0152] Although preferred embodiments of the present invention have been described, those skilled in the art, upon learning the basic inventive concept, can make other changes and modifications to these embodiments. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments as well as all changes and modifications falling within the scope of the embodiments of the present invention.

[0153] Finally, it should be noted that in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or terminal device that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or terminal device. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or terminal device that includes said element.

[0154] The present invention provides a detailed description of an anomaly monitoring system, method, device, electronic device, and readable storage medium for smart homes. Specific examples have been used to illustrate the principles and implementation methods of the present invention. The descriptions of the above embodiments are only for the purpose of helping to understand the method and core ideas of the present invention. At the same time, those skilled in the art will recognize that, based on the ideas of the present invention, there will be changes in the specific implementation methods and application scope. Therefore, the content of this specification should not be construed as a limitation of the present invention.

Claims

1. An anomaly monitoring system for smart homes, characterized in that, The anomaly monitoring system includes multiple monitoring devices, target devices, and gateway devices; The monitoring device is used to collect physical data of the target device and send it to the gateway device; receive fused data sent by the gateway device, generate status information based on the fused data and send it to the gateway device; The gateway device is configured to perform fusion processing on physical data from multiple monitoring devices to obtain fused data; send the fused data to multiple monitoring devices; and make a comprehensive decision based on the received status information from multiple monitoring devices to obtain an anomaly judgment result.

2. The smart home anomaly monitoring system according to claim 1, characterized in that, The gateway device is further configured to assign monitoring tasks to the monitoring device and send them to the monitoring device; the monitoring task is used to instruct the monitoring device to collect data from the corresponding target device. The monitoring device is also used to receive the monitoring task, determine the corresponding target device according to the monitoring task, and collect the physical data of the target device.

3. The smart home anomaly monitoring system according to claim 2, characterized in that, The gateway device is specifically used to acquire historical acquisition quality information of the monitoring device and location information between the monitoring device and the target device; and to determine the monitoring task of the monitoring device based on the historical acquisition quality information and the location information.

4. The anomaly monitoring system for smart homes according to claim 1, characterized in that, The physical data includes acoustic fingerprint data and / or vibration data; the monitoring equipment is equipped with an acoustic fingerprint acquisition device and / or a vibration acquisition device. The monitoring equipment is specifically used to collect the acoustic data of the target device through the acoustic data acquisition device; and / or to collect the vibration data of the target device through the vibration acquisition device.

5. The anomaly monitoring system for smart homes according to claim 3, characterized in that, The gateway device is specifically used to determine the weight of the corresponding monitoring device based on the historical quality information and the location information; and to perform weighted processing on the physical data collected by the monitoring device based on the weight of the monitoring device to obtain the fused data.

6. The anomaly monitoring system for smart homes according to claim 1, characterized in that, A preset first diagnostic model is deployed on the monitoring equipment; The monitoring device is specifically used to extract features from the fused data to obtain feature information; The feature information is input into the first diagnostic model to obtain the status information output by the first diagnostic model.

7. The anomaly monitoring system for smart homes according to claim 6, characterized in that, A preset second diagnostic model is deployed on the gateway device; The gateway device is further configured to acquire the feature information when the abnormal diagnosis conditions are met; input the feature information into the second diagnostic model to obtain the abnormal type and solution output by the second diagnostic model.

8. The anomaly monitoring system for smart homes according to claim 7, characterized in that, The monitoring device is also used to train the first diagnostic model using the physical data to obtain model parameter update information; encrypt the model parameter update information and send it to the gateway device; The first diagnostic model is updated based on the model update knowledge sent by the gateway device; The gateway device is further configured to update the second diagnostic model based on encrypted model parameter update information from multiple monitoring devices, obtain model update knowledge, and send it to the monitoring devices.

9. A method for detecting anomalies in smart homes, characterized in that, The method, applied to monitoring equipment, includes: The physical data of the target device is collected and sent to the gateway device. The physical data is used by the gateway device to generate fused data. The system receives fused data sent by the gateway device, generates status information based on the fused data, and sends it to the gateway device. The status information is used by the gateway device to make comprehensive decisions to obtain anomaly judgment results.

10. The anomaly detection method for smart homes according to claim 9, characterized in that, The method further includes: The monitoring device receives a monitoring task, determines the corresponding target device based on the monitoring task, and collects physical data of the target device. The monitoring task is assigned by the gateway device and is used to instruct the monitoring device to collect data from the corresponding target device.

11. The anomaly detection method for smart homes according to claim 9, characterized in that, The physical data includes acoustic fingerprint data and / or vibration data; the monitoring equipment is equipped with an acoustic fingerprint acquisition device and / or a vibration acquisition device. The physical data of the target device being collected includes: The voiceprint data of the target device is collected using the voiceprint acquisition device. and / or Vibration data of the target equipment are collected using the vibration acquisition device.

12. The anomaly detection method for smart homes according to claim 9, characterized in that, A preset first diagnostic model is deployed on the monitoring equipment; The step of generating status information based on the fusion data sent by the gateway device includes: Feature extraction is performed on the fused data to obtain feature information; The feature information is input into the first diagnostic model to obtain the status information output by the first diagnostic model.

13. The anomaly monitoring method for smart homes according to claim 12, characterized in that, A preset second diagnostic model is deployed on the gateway device; The method further includes: The first diagnostic model is trained using the physical data to obtain model parameter update information; The model parameter update information is encrypted and sent to the gateway device; the encrypted model parameter update information is used by the gateway device to update the second diagnostic model to obtain model update knowledge; The first diagnostic model is updated based on the model update knowledge sent by the gateway device.

14. A method for detecting anomalies in smart homes, characterized in that, Applied to a gateway device, the method includes: Physical data from multiple monitoring devices are fused to obtain fused data; the physical data is collected by the monitoring devices from the target device. The fused data is sent to multiple monitoring devices; the fused data is used by the monitoring devices to generate status information. An anomaly judgment result is obtained by comprehensively making decisions based on the status information received from multiple monitoring devices.

15. The anomaly detection method for smart homes according to claim 14, characterized in that, The method further includes: A monitoring task is assigned to the monitoring device and sent to the monitoring device; the monitoring task is used to instruct the monitoring device to collect data from the corresponding target device.

16. The anomaly detection method for smart homes according to claim 15, characterized in that, Assigning monitoring tasks to the monitoring equipment includes: Acquire historical data acquisition quality information of the monitoring device and location information between the monitoring device and the target device; The monitoring task of the monitoring equipment is determined based on the historical quality information and the location information.

17. The anomaly detection method for smart homes according to claim 16, characterized in that, The process of fusing physical data from multiple monitoring devices to obtain the fused data includes: The weight of the corresponding monitoring device is determined based on the historical data collection quality information and the location information. The physical data collected by the monitoring devices are weighted according to their respective weights to obtain the fused data.

18. The anomaly detection method for smart homes according to claim 14, characterized in that, A preset second diagnostic model is deployed on the gateway device; The method further includes: When the abnormal diagnostic conditions are met, feature information is acquired; the feature information is obtained by the monitoring device extracting features from the fused data. The feature information is input into the second diagnostic model to obtain the abnormality type and solution output by the second diagnostic model.

19. The anomaly detection method for smart homes according to claim 18, characterized in that, A preset first diagnostic model is deployed on the monitoring equipment; The method further includes: The second diagnostic model is updated based on encrypted model parameter update information from multiple monitoring devices to obtain model update knowledge, which is then sent to the monitoring devices. The model parameter update information is obtained by the monitoring device through training the first diagnostic model using the physical data; the model update knowledge is used by the monitoring device to update the first diagnostic model.

20. An anomaly monitoring device for smart homes, characterized in that, Applied to monitoring equipment, the device includes: The physical data acquisition module is used to acquire physical data of the target device and send it to the gateway device. The physical data is used by the gateway device to generate fused data. The status information generation module is used to receive the fused data sent by the gateway device, generate status information based on the fused data, and send it to the gateway device; the status information is used by the gateway device for comprehensive processing to obtain an anomaly judgment result.

21. An anomaly monitoring device for smart homes, characterized in that, Applied to a gateway device, the device includes: The fusion data generation module is used to fuse physical data from multiple monitoring devices to obtain fused data; the physical data is collected by the monitoring devices from the target device. A fused data transmission module is used to transmit the fused data to multiple monitoring devices; the fused data is used by the monitoring devices to generate status information. The anomaly judgment result determination module is used to make a comprehensive decision based on the received status information from multiple monitoring devices to obtain an anomaly judgment result.

22. An electronic device, characterized in that, It includes a processor, a memory, and a program or instructions stored in the memory and executable on the processor, wherein the program or instructions, when executed by the processor, implement the steps of the anomaly monitoring method for smart homes as described in claims 9-13 or 14-19.

23. A readable storage medium, characterized in that, The readable storage medium stores a program or instructions that, when executed by a processor, implement the steps of anomaly monitoring in a smart home as described in claims 9-13 or 14-19.

Images