A method and system for managing smart home devices

Abstract

The present invention discloses a smart home device management method and system, which relates to the technical fields of data service value evaluation and decision optimization, and includes: collecting device data and fusing it to form a unified device status data set; based on the device status data set, performing feature extraction and combining to generate a multi-dimensional life prediction feature vector; according to the multi-dimensional life prediction feature vector, constructing a device life prediction model and real-time outputting a predicted value of the remaining life of the device; based on the predicted value of the remaining life of the device, identifying abnormal decreases in the device life and triggering device maintenance warning information; according to the device maintenance warning information, receiving user feedback and adaptively adjusting the parameters of the device life prediction model. The present invention fuses multi-features such as vibration and thermoelectricity to construct a life model, updates adaptively and gives hierarchical warnings, realizes early maintenance of household appliances, reduces the risk of fault shutdown, saves operation and maintenance costs, and has significant benefits in optimizing energy consumption monitoring, greatly improving the user experience.

Description

Technical Field

[0001] The present invention relates to the technical field of smart home device management and health prediction modeling, and specifically to a smart home device management method and system. Background Art

[0002] With the continuous development of Internet of Things (IoT) technology and artificial intelligence (AI), smart home systems have evolved from the initial stage of remote device control to the current comprehensive management stage with multi-device collaboration, status awareness, and intelligent decision-making in parallel. More and more home terminal devices (such as smart air conditioners, smart lighting, security cameras, etc.) have achieved networking and data collection functions, providing users with convenient and comfortable living experiences. At the same time, the types and quantities of smart devices continue to increase, significantly raising the complexity of device management. To improve device operation efficiency and service life, device health management technology has gradually been introduced into the home scenario, forming a device life prediction and fault warning mechanism based on data analysis. However, most existing methods focus on a single device, lacking cross-device status fusion and systematic management means, and unable to fully unleash the data potential of smart home systems.

[0003] Although some studies in the prior art have attempted to introduce data-driven health assessment and prediction models, there are still multiple key technical bottlenecks. First, traditional methods usually rely on single-dimensional or static features and lack the ability to fuse multi-dimensional and time-series features, resulting in low accuracy of life prediction models. Second, existing device status monitoring means are mostly based on regular maintenance or post-fault processing, failing to achieve early warning and proactive maintenance. In addition, existing models generally adopt a fixed parameter structure and lack a mechanism for dynamic adaptive optimization based on actual operation feedback, unable to effectively cope with actual environmental changes such as user behavior differences and different degrees of device aging. These technical limitations make existing device management methods unable to achieve efficient and personalized device health management and intelligent warning, and it is also difficult to build a continuously evolving and feedback-driven management system. The present invention constructs a unified data fusion framework, a multi-dimensional feature extraction mechanism, and a prediction model with self-learning ability to achieve high-precision prediction of device status and proactive maintenance suggestions in a multi-device environment, breaking through the limitations of the prior art in data dimension fusion, model evolution ability, and warning accuracy, and having significant practical value and innovative advantages. Summary of the Invention

[0004] In view of the above problems, the present invention is proposed.

[0005] Therefore, the technical problem to be solved by the present invention is that the existing smart home device management methods lack a multi-device status fusion mechanism, have low accuracy in the life prediction model and are difficult to dynamically optimize, the device maintenance and feedback mechanism is imperfect and it is difficult to form a closed-loop management, and how to dynamically adjust the parameters of the life prediction model based on user feedback to achieve the adaptive evolution of the prediction model.

[0006] To solve the above technical problems, the present invention provides the following technical solutions:

[0007] In the first aspect, an embodiment of the present invention provides a smart home device management method, including collecting device data and fusing it to form a unified device status data set;

[0008] Based on the device status data set, feature extraction is performed to combine and generate a multi-dimensional life prediction feature vector;

[0009] According to the multi-dimensional life prediction feature vector, a device life prediction model is constructed to output the predicted remaining life value of the device in real time;

[0010] Based on the predicted remaining life value of the device, identify the abnormal decline in the device life and trigger a device maintenance warning message;

[0011] According to the device maintenance warning message, receive user feedback and adaptively adjust the parameters of the device life prediction model.

[0012] As a preferred solution of the smart home device management method of the present invention, where: the collecting device data includes respectively deploying data collection units in each device in the smart home environment, and the data collection unit includes a vibration sensor, a temperature sensor, an acceleration sensor, a current sensor and a power sensor;

[0013] The vibration sensor is used to collect the vibration signal during the operation of the device and generate a multi-band vibration envelope energy sequence through multi-band envelope analysis;

[0014] The temperature sensor is used to collect the temperature change of the key components of the device in real time to form a temperature time series curve;

[0015] The acceleration sensor is used to measure the acceleration change of the device and obtain the speed change rate time series through numerical calculation;

[0016] The current sensor is used to measure the current change during the operation of the device in real time to generate current waveform data;

[0017] The power sensor is used to measure the power output of the device in real time to generate instantaneous power data;

[0018] The data acquisition unit periodically sends the multi - band vibration envelope energy sequence, temperature time - series curve, speed change rate time - series, current waveform data, and instantaneous power data to the smart home local gateway through the local wireless communication device for data storage. The smart home local gateway performs preliminary time - marking processing on the received data to form the original device operation data with a unified timestamp.

[0019] As a preferred solution of the smart home device management method described in the present invention, wherein: the fusion to form a unified device status data set includes, based on the original device operation data stored in the smart home local gateway, using a time - series data interpolation algorithm to eliminate the difference in data sampling frequencies of different devices and unify the data sampling frequency;

[0020] Standardize the original device operation data through a data standardization algorithm to eliminate the differences in dimension and value range between data, and generate standardized device operation data;

[0021] Use a multi - source data spatio - temporal alignment algorithm for the standardized device operation data. With a unified device identifier and a unified timestamp as the data matching benchmarks, integrate the data item - by - item according to the timestamp to form spatio - temporally consistent multi - dimensional device operation data;

[0022] Perform data noise reduction processing on the multi - dimensional device operation data using the Kalman filtering algorithm to eliminate the measurement errors of various sensors and environmental noise interference, and generate noise - reduced multi - dimensional device operation data;

[0023] Fuse the noise - reduced multi - dimensional device operation data according to the preset weights through a multi - dimensional data weighted fusion algorithm to obtain a unified data format and output it as a unified device status data set;

[0024] The device status data set includes the multi - band vibration envelope energy sequence after data processing, the temperature time - series curve after data processing, the speed change rate time - series after data processing, the current waveform data after data processing, and the instantaneous power data after data processing.

[0025] As a preferred solution of the smart home device management method described in the present invention, wherein: the generation of the multi - dimensional life prediction feature vector includes, performing weighted summation on the multi - band vibration envelope energy sequence after data processing according to the preset weights to obtain the vibration fatigue feature;

[0026] Perform integral processing on the temperature time - series curve after data processing to obtain the cumulative heat load feature;

[0027] Perform square integral on the speed change rate time - series after data processing to obtain the speed excitation feature;

[0028] Accumulate the absolute differences of adjacent sampling points of the current waveform data after data processing to obtain the current disturbance feature;

[0029] Perform logarithmic mapping and normalization on the instantaneous power data after data processing to obtain the power response feature;

[0030] Combine the vibration fatigue feature, cumulative heat load feature, speed excitation feature, current disturbance feature and power response feature in a preset order to form a multi-dimensional life prediction feature vector.

[0031] As a preferred solution of the smart home device management method described in the present invention, wherein: constructing the device life prediction model includes inputting the multi-dimensional life prediction feature vector into a preset life prediction function group, and the life prediction function group includes a non-linear mapping function for vibration fatigue response features, an incomplete integration function for cumulative heat load features and speed excitation features, a logarithmic normalization function for power response features, and a radical attenuation function for adjusting the influence degree of current disturbance features;

[0032] Perform weighted combination on the output results of the life prediction function group, perform integral accumulation, and form the health loss value of the device under the current operating state;

[0033] Combine the state disturbance complexity factor during the device operation, perform normalization processing on the health loss value, and output the device remaining life prediction value representing the remaining life status of the device.

[0034] As a preferred solution of the smart home device management method described in the present invention, wherein: identifying the abnormal decline of the device life includes constructing a dynamic health index baseline for the device health level according to the type of the target device and historical operation data;

[0035] Periodically calculate the residual difference between the current device remaining life prediction value and the dynamic health index baseline, and use a sliding time window to accumulate the residuals. If the continuous period exceeds the set threshold and the cumulative residual value exceeds the set threshold, it is determined that there is an abnormal decline trend in life;

[0036] Divide multiple levels of maintenance warning levels according to the residual degree, including primary warning, intermediate warning and emergency warning. The primary warning is used to trigger device self-check and user prompt, the intermediate warning is used to limit the device operation ability and record abnormal energy consumption, and the emergency warning is used to automatically cut off the power of high-risk devices and push remote maintenance notifications;

[0037] After the warning level is confirmed, automatically generate device maintenance warning information, and the device maintenance warning information includes the prediction value change trend, the current level classification and recommended treatment measures.

[0038] As a preferred solution of the smart home device management method described in the present invention, wherein: the adaptive adjustment of the parameters of the device life prediction model includes, after the device maintenance warning information is pushed, receiving feedback information provided by the user through the smart home terminal, and the feedback information includes the maintenance execution situation confirmed by the user, the actual operation state of the device, the user's subjective evaluation and other relevant information;

[0039] Compare the feedback information with the predicted value of the remaining device life. If it is found that there is a deviation between the predicted value and the actual device state, an adaptive adjustment algorithm is used to optimize the parameters of the life prediction model.

[0040] In a second aspect, an embodiment of the present invention provides a smart home device management system, including:

[0041] Data acquisition module: Collect device data and integrate it to form a unified set of device status data;

[0042] Feature extraction module: Based on the set of device status data, perform feature extraction and combine to generate a multi-dimensional life prediction feature vector;

[0043] Life prediction module: Based on the multi-dimensional life prediction feature vector, construct a device life prediction model and output the predicted value of the remaining device life in real time;

[0044] Abnormality recognition and warning module: Based on the predicted value of the remaining device life, recognize the situation of abnormal decline in device life and trigger device maintenance warning information;

[0045] Feedback learning module: According to the device maintenance warning information, receive user feedback and adaptively adjust the parameters of the device life prediction model.

[0046] Advantages of the present invention: By constructing a unified set of device status data, combining multi-dimensional feature vectors and life prediction models, the present invention can achieve high-precision real-time evaluation of the operation status of multiple types of devices in the smart home and prediction of the remaining life. The multi-level maintenance warning mechanism triggered based on the prediction results effectively improves the initiative and safety of device maintenance and avoids interference or losses caused by sudden device failures to users. At the same time, the present invention introduces a user feedback-driven adaptive parameter optimization mechanism, enabling the life prediction model to have the ability of dynamic adjustment and personalized evolution, significantly enhancing the adaptability and robustness of the system, constructing an intelligent management closed-loop from data acquisition, status evaluation, warning trigger to feedback correction, and comprehensively improving the management efficiency and usage experience of smart home devices. Description of the Drawings

[0047] To more clearly illustrate the technical solutions of the embodiments of the present invention, the following will briefly introduce the drawings required for the description of the embodiments. Obviously, the drawings in the following description are only some embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other drawings can be obtained based on these drawings, where:

[0048] Figure 1 It is the overall flowchart of a smart home device management method provided by the first embodiment of the present invention. Specific implementation manners

[0049] To make the above objects, features, and advantages of the present invention more obvious and understandable, the following will make a detailed description of the specific implementation manners of the present invention with reference to the drawings of the specification. Obviously, the described embodiments are some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

[0050] Embodiment 1, referring to Figure 1 This is an embodiment of the present invention, providing a smart home device management method, including:

[0051] S1: Collect device data and fuse it to form a unified set of device status data.

[0052] S11: Deploy data collection units in each device in the smart home environment. The data collection unit includes a vibration sensor, a temperature sensor, an acceleration sensor, a current sensor, and a power sensor;

[0053] The vibration sensor is used to collect vibration signals during the operation of the device and generate a multi-band vibration envelope energy sequence through multi-band envelope analysis;

[0054] The temperature sensor is used to collect the temperature changes of the key components of the device in real time and form a temperature time series curve;

[0055] The acceleration sensor is used to measure the acceleration changes of the device and obtain the speed change rate time series through numerical calculation;

[0056] The current sensor is used to measure the current changes during the operation of the device in real time and generate current waveform data;

[0057] The power sensor is used to measure the power output of the device in real time and generate instantaneous power data;

[0058] The data acquisition unit periodically sends the multi-band vibration envelope energy sequence, temperature time series curve, speed change rate time series, current waveform data, and instantaneous power data to the smart home local gateway through the local wireless communication device for data storage. The smart home local gateway performs preliminary time marking processing on the received data to form the original device operation data with a unified timestamp.

[0059] Specifically, the vibration sensor is used to collect the vibration signal during the device operation and perform multi-band envelope analysis to generate the multi-band vibration envelope energy sequence. This sequence can effectively capture the vibration characteristics of the device at different frequencies, especially the subtle changes in the high-frequency band, which helps to identify problems such as fatigue damage or looseness of mechanical components at an early stage.

[0060] The temperature sensor continuously collects the temperature changes of the key components of the device to form the temperature time series curve. This curve reflects the thermal load of the device under different working conditions and can reveal potential problems such as overheating and poor heat dissipation of the device.

[0061] The acceleration sensor is used to measure the acceleration change of the device and perform numerical calculation to obtain the speed change rate time series. This time series data can reflect the mechanical shock load during the dynamic processes such as startup and stop of the device, which helps to evaluate the fatigue degree of the device during frequent start-stop processes.

[0062] The current sensor and power sensor are respectively used to continuously measure the current change and power output during the device operation to generate the current waveform data and instantaneous power data. These data can reveal the electrical performance changes of the device under different load conditions and help to identify abnormalities in the electrical system, such as overload and short circuit.

[0063] The data acquisition unit periodically sends the above-mentioned various data to the smart home local gateway through the local wireless communication device for data storage and performs preliminary time marking processing to form the original device operation data with a unified timestamp. This unified data format helps with subsequent data fusion and analysis, improving the efficiency and accuracy of data processing.

[0064] S12: Based on the original device operation data stored in the smart home local gateway, use the time series data interpolation algorithm to eliminate the differences in data sampling frequencies of different devices and unify the data sampling frequency;

[0065] Standardize the original device operation data through the data standardization algorithm to eliminate the differences in dimension and value range between data and generate the standardized device operation data;

[0066] Use the multi-source data spatio-temporal alignment algorithm for the standardized data of device operation. With the unified device identifier and unified timestamp as the data matching benchmarks, integrate the data item by item according to the timestamp to form multi-dimensional device operation data that is spatio-temporally consistent.

[0067] Use the Kalman filtering algorithm to perform data denoising on the multi-dimensional device operation data, eliminate the measurement errors of various sensors and environmental noise interference, and generate the denoised multi-dimensional device operation data.

[0068] Fuse the denoised multi-dimensional device operation data according to the preset weights through the multi-dimensional data weighted fusion algorithm to obtain a unified data format and output it as a unified set of device status data.

[0069] The set of device status data includes the multi-band vibration envelope energy sequence after data processing, the temperature time series curve after data processing, the time series of the speed change rate after data processing, the current waveform data after data processing, and the instantaneous power data after data processing.

[0070] It should be noted that in the smart home environment, sensors of different devices may have different sampling frequencies, resulting in inconsistencies in data on the time axis. By using the time series data interpolation algorithm, such as linear interpolation or spline interpolation, data with different frequencies can be unified to a standard sampling frequency to ensure the temporal consistency of subsequent processing.

[0071] Since the data collected by various sensors have different dimensions and value ranges, such as temperature (°C), current (A), power (W), etc., directly performing data fusion may cause the influence of some features on the model to be amplified or reduced. Through standardization processing, such as Z-score standardization or Min-Max normalization, the dimension differences can be eliminated, enabling each feature to be on the same scale, which is convenient for subsequent analysis and modeling.

[0072] In the data collection of multiple devices and multiple sensors, there may be problems such as inconsistent timestamps or confused device identifiers. By unifying the device identifier and timestamp and using the spatio-temporal alignment algorithm, it is possible to ensure the consistency of data from different sources in terms of time and space, and form complete multi-dimensional device operation data.

[0073] Sensors may be affected by environmental noise or their own errors during data collection, resulting in noise in the data. Kalman filtering is an effective denoising method that can filter out noise through the processes of prediction and update to obtain more accurate device operation status data.

[0074] After obtaining the noise reduction data of multiple sensors, it is necessary to fuse them into a unified set of device status data. By setting the preset weights of the data of each sensor and adopting a weighted fusion algorithm, the characteristics of various types of data can be integrated to form a unified data format, providing a reliable data basis for subsequent feature extraction and model construction.

[0075] S2: Based on the set of device status data, perform feature extraction and combine to generate a multi-dimensional life prediction feature vector.

[0076] Perform weighted summation on the multi-band vibration envelope energy sequence obtained by processing the data in the set of device status data according to the preset weights to obtain the vibration fatigue feature; specifically, process the multi-band vibration envelope energy sequence. This sequence is obtained by performing multi-band envelope analysis on the vibration signal during the operation of the device and reflects the vibration energy distribution of the device in different frequency ranges. In order to extract the vibration fatigue feature, it is necessary to perform weighted summation on the energy values of each frequency band according to the preset weights. The setting of the weights should be based on the sensitivity of each frequency band to the fatigue damage of the device and is usually determined through experiments or historical data analysis. The result of the weighted summation is the vibration fatigue feature, which can effectively reflect the influence of the vibration energy of the device in each frequency band on its life.

[0077] Perform integral processing on the temperature time series curve after data processing to obtain the cumulative heat load feature; specifically, process the temperature time series curve. This curve records the temperature change of the key components of the device during operation. By performing integral processing on this curve, the cumulative heat load feature of the device within a certain period of time can be obtained. This feature reflects the cumulative degree of thermal stress endured by the device during operation and is of great significance for evaluating the thermal aging and life of the device.

[0078] Perform square integral on the time series of the rate of change of speed after data processing to obtain the speed excitation feature; specifically, process the time series of the rate of change of speed. This time series data is obtained by performing numerical calculations on the acceleration signal of the device and reflects the dynamic characteristics of the speed change of the device. By performing square integral processing on the time series of the rate of change of speed, the speed excitation feature can be obtained. This feature can quantify the intensity of the speed change experienced by the device during operation and plays an important role in evaluating the mechanical shock and fatigue damage of the device.

[0079] Accumulate the absolute values of the differences between adjacent sampling points of the current waveform data after data processing to obtain the current disturbance feature; specifically, this data records the current change of the device during operation. By calculating the absolute values of the differences between adjacent sampling points and accumulating them, the current disturbance feature can be obtained. This feature reflects the degree of current fluctuation of the device during operation and is of great significance for evaluating the electrical stability and potential faults of the device.

[0080] Perform logarithmic mapping and normalization on the instantaneous power data after data processing to obtain power response characteristics; specifically, process the instantaneous power data. This data records the power output of the device during operation. By performing logarithmic mapping and normalization on this data, power response characteristics can be obtained. These characteristics can reflect the power response characteristics of the device under different load conditions and play an important role in evaluating the energy efficiency and operating status of the device.

[0081] Combine the vibration fatigue characteristics, cumulative heat load characteristics, speed excitation characteristics, current perturbation characteristics, and power response characteristics in a preset order to form a multi-dimensional life prediction feature vector.

[0082] S3: Construct a device life prediction model based on the multi-dimensional life prediction feature vector and output the predicted value of the remaining life of the device in real time.

[0083] Input the multi-dimensional life prediction feature vector into a preset set of life prediction functions. The set of life prediction functions includes a non-linear mapping function for vibration fatigue response characteristics, an incomplete integration function for cumulative heat load characteristics and speed excitation characteristics, a logarithmic normalization function for power response characteristics, and a radical decay function for adjusting the influence degree of current perturbation characteristics;

[0084] Specifically:

[0085] Vibration fatigue characteristics Obtained through wavelet packet decomposition and energy weighting:

[0086] ;

[0087] Where is the vibration fatigue characteristic at time , is the th frequency band's vibration envelope energy at time ; is the vibration weight of the th frequency band, representing its diagnostic importance or feature sensitivity, and is generally obtained by empirical determination or weighted coefficient based on entropy weight; is the frequency band index number, with a value range of ; is the total number of wavelet packet frequency bands;

[0088] The non-linear fatigue mapping function is defined as:

[0089] ;

[0090] Where is the non-linear mapping vibration energy response, constituting the fatigue index; is The squared value is used to construct a non - linear fatigue index;

[0091] The device cumulative heat load characteristic and the speed excitation characteristic are respectively defined as:

[0092] ;

[0093] Among them, is the cumulative integral value within the interval of temperature (cumulative heat load characteristic), reflecting the heat load; is the temperature at time point ; is the temperature at time point ; is the rate of change of speed at time point is its squared value, used to evaluate the contribution of mechanical excitation energy, indicating the impact of shock or mutation on life; is the squared integral value of the rate of change of speed within the interval of (speed excitation characteristic);

[0094] The non - linear life response function is expressed in the form of an incomplete Gamma function as:

[0095] ;

[0096] Among them, respectively represent the cumulative amounts of heat load and speed change; is the incomplete Gamma function, used to express non - linear damage accumulation; is the integration variable, which is an intermediate variable used to traverse the domain in the mathematical integration process;

[0097] Meanwhile, a power response adjustment term is introduced:

[0098] ;

[0099] Among them, is a power - based normalization adjustment function, used to adjust the loss growth in the high - power interval; is the instantaneous power at time (power response characteristic); is the normalization scaling coefficient, controlling the "compression" degree of the adjustment function for the high - power interval, the larger it is, the stronger the "adjustment" for high power, making the influence of the instantaneous power on gradually weaken, thus reflecting the adaptive characteristics of the system to power influence.

[0100] constitutes a heat - motion - power composite fatigue mapping function:

[0101] ;

[0102] Among them, is the damage intensity value under thermal-velocity-power coupling;

[0103] During the sampling period, calculate the current fluctuation characteristics:

[0104] ;

[0105] Among them, is the quantization value of the current disturbance intensity per unit time (current disturbance characteristic); is time At the moment, the current value at the th sampling point; is the sampling point index, ; is the power coefficient of the disturbance intensity, reflecting the sensitivity of the system to the current fluctuation amplitude. The larger it is, the more sensitive it is to large fluctuations, mainly used to enhance the response ability to spike-type disturbances or nonlinear changes, and plays a role of nonlinear amplification or suppression when measuring the "current disturbance characteristic".

[0106] Weight the output results of the life prediction function group, perform integral accumulation, and form the health loss value of the device under the current operating state ; Specifically:

[0107] ;

[0108] Combine the state disturbance complexity factor during the device operation, normalize the health loss value, and output the device remaining life prediction value representing the remaining life status of the device ; Specifically:

[0109] ;

[0110] Among them, is the device remaining life prediction value of the device within the time window , as the core output value of life assessment; is the evaluation time window length, is the normalization adjustment factor; is the state disturbance complexity normalization term, defined as:

[0111] ;

[0112] Among them, is the device operation state switching times index, ; represents the The current change rate before and after the secondary state transition is the load difference corresponding to the secondary state transition is the coupling sensitivity coefficient of temperature rise and jump signal in the secondary state transition.

[0113] S4: Based on the predicted value of the remaining life of the device, identify the abnormal decline of the device life, and trigger the device maintenance warning information.

[0114] According to the type and historical operation data of the target device, construct a dynamic health index baseline for the device health level, and based on the actual operation duration of the device, generate a theoretical life decay curve in combination with the Weibull distribution model as the reference for judging the remaining life;

[0115] Periodically calculate the residual difference between the current predicted value of the remaining life of the device and the dynamic health index baseline, and use a sliding time window to accumulate the residuals. If the continuous period exceeds the set threshold and the cumulative residual value exceeds the set threshold, it is judged that there is an abnormal downward trend in life;

[0116] Set the residual tolerance coefficient according to the structural sensitivity of different types of devices, relax the threshold for high-vibration devices, and tighten the threshold for precision devices to adapt to the performance degradation characteristics of multiple types of devices;

[0117] Divide multiple levels of maintenance warning levels according to the degree of residuals, including primary warning, intermediate warning and emergency warning. The primary warning is used to trigger device self-check and user prompt, the intermediate warning is used to limit the device operation ability and record abnormal energy consumption, and the emergency warning is used to automatically cut off the power of high-risk devices and push remote maintenance notifications;

[0118] After the warning level is confirmed, automatically generate the device maintenance warning information, and the device maintenance warning information includes the predicted value change trend, the current level classification and the recommended treatment measures.

[0119] It should be noted that the "dynamic health index baseline" is not a static threshold setting, but a baseline model established based on the operating history and type differences of the target device, which reflects the dynamic evolution of the device health status and avoids misjudgment problems under fixed thresholds. The "Weibull distribution model", as a life statistical model widely used in reliability analysis, can more accurately reflect the probability distribution characteristics of the life attenuation of most devices. Its rationality lies in its fitting advantage for right-skewed life data, enhancing the rigor of theoretical life judgment. The "residual difference" represents the degree of deviation between the predicted value and the health baseline, and is a sensitive indicator for measuring sudden changes in device status; its accumulation method is processed through a "sliding time window", so that short-term fluctuations will not accidentally trigger an alarm, while retaining trend information and enhancing judgment robustness. Setting the "residual tolerance coefficient" reflects the adaptive design of the algorithm, that is, tolerance correction is introduced for the high-frequency fluctuations that may exist in high-vibration devices, reflecting the model generalization ability; for precision devices, higher discrimination sensitivity is required to improve monitoring accuracy. The division of the "multi-level maintenance warning level" enables the device maintenance strategy to have response levels, which can not only intervene in advance, but also respond according to the risk progression, helping to achieve refined device management.

[0120] S5: According to the device maintenance warning information, receive user feedback and adaptively adjust the parameters of the device life prediction model.

[0121] After the device maintenance warning information is pushed, receive the feedback information provided by the user through the smart home terminal. The feedback information includes the maintenance execution situation confirmed by the user, the actual operating status of the device, the user's subjective evaluation and other relevant information.

[0122] Compare the feedback information with the predicted value of the remaining life of the device. If a deviation is found between the predicted value and the actual device status, use an adaptive adjustment algorithm to optimize the parameters of the life prediction model.

[0123] It should be noted that after the device maintenance warning information is pushed, the system not only receives the active feedback information from the user, such as the maintenance execution situation confirmed by the user through the smart home terminal, but also includes the passive monitoring data of the device operating status and unstructured information such as the user's subjective evaluation. These feedback data are diverse in type and different in structure, but together they constitute an important information source for the true operating status of the device and are the core basis for verifying the accuracy of device life prediction and adjusting model parameters.

[0124] Furthermore, the system compares the above feedback information with the remaining life prediction value generated by the current life prediction model, and automatically judges the accuracy of the model prediction based on the error value between the actual operating state and the predicted data. If there is a systematic deviation between the actual state indicated by the feedback information and the model prediction in multiple consecutive cycles, the system will trigger an error-driven adaptive optimization algorithm to fine-tune the key parameters in the life model. The adjusted parameters can involve the scale and shape parameters in the life distribution model, or can also include model environment parameters such as the sliding window size, residual threshold, and tolerance coefficient for state recognition, so as to achieve a more accurate match of the degradation characteristics of different devices.

[0125] Moreover, the system conducts intelligent classification and weight assignment for the type and quality of the feedback information. Contents such as whether the equipment returns to normal after the user confirms maintenance, whether new abnormal signs appear, and the running smoothness in the user's subjective perception are all included in the evaluation scope. For voice or text-based feedback information, a natural language processing model can also be introduced to extract keywords and recognize emotions, and convert them into parameters with semantic dimensions to assist the model correction process.

[0126] In terms of the model parameter adjustment mechanism, the system can adopt a periodic training method. By constructing a sample set with historical feedback data, global or local search for model parameters is carried out using technologies such as genetic algorithms, Bayesian updates, or gradient optimization, and finally a prediction model with better convergence and stronger generalization ability is output. Compared with the static model, such methods can significantly improve the application stability and accuracy in the actual environment.

[0127] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit them. Although the present invention has been described in detail with reference to the preferred embodiments, those of ordinary skill in the art should understand that the technical solutions of the present invention can be modified or equivalently replaced without departing from the spirit and scope of the technical solutions of the present invention, and they should all be covered within the scope of the claims of the present invention.

[0128] Embodiment 2 is the second embodiment of the present invention. What is different from the previous embodiment is:

[0129] When the above-mentioned functions are implemented in the form of software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art or the part of the current technical solution can be embodied in the form of a software product. The current computer software product is stored in a storage medium and includes several instructions for causing a computer device (which may be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods described in various embodiments of the present invention. The aforementioned storage medium includes: USB flash drives, mobile hard disks, read-only memories (ROM, Read-Only Memory), random access memories (RAM, Random Access Memory), magnetic disks, or optical discs, etc., all kinds of media that can store program codes.

[0130] The logic and / or steps represented in the flowchart or described in other ways herein, for example, can be considered as a definite sequence list of executable instructions for implementing logical functions, and can be specifically implemented in any computer-readable medium for use by an instruction execution system, apparatus, or device (such as a computer-based system, a system including a processor, or other systems that can fetch instructions from the instruction execution system, apparatus, or device and execute the instructions), or in combination with these instruction execution systems, apparatus, or devices. For the purposes of this specification, a "computer-readable medium" can be any device that can contain, store, communicate, propagate, or transport a program for use by or in combination with an instruction execution system, apparatus, or device.

[0131] More specific examples (non-exhaustive list) of computer-readable media include the following: electrical connection parts with one or more wirings (electronic devices), portable computer disk cartridges (magnetic devices), random access memories (RAM), read-only memories (ROM), erasable programmable read-only memories (EPROM or flash memories), fiber optic devices, and portable compact disc read-only memories (CDROM). Additionally, the computer-readable medium can even be paper or other suitable media on which the program can be printed, because the program can be obtained electronically, for example, by optically scanning the paper or other media, then editing, interpreting, or processing it in other suitable ways as necessary, and then storing it in a computer memory.

[0132] Embodiment 3, an embodiment of the present invention, provides a smart home device management system, including a data acquisition module, a feature extraction module, a life prediction module, an anomaly identification and warning module, and a feedback learning module.

[0133] Data acquisition module: Collect device data and fuse it to form a unified set of device status data;

[0134] Feature extraction module: Based on the set of device status data, perform feature extraction and combine to generate a multi-dimensional life prediction feature vector;

[0135] Life prediction module: Based on the multi-dimensional life prediction feature vector, construct a device life prediction model and real-time output the predicted value of the remaining life of the device;

[0136] Abnormality identification and warning module: Based on the predicted value of the remaining life of the device, identify the abnormal decrease in the device life and trigger the device maintenance warning information;

[0137] Feedback learning module: According to the device maintenance warning information, receive user feedback and adaptively adjust the parameters of the device life prediction model.

Claims

1. A method for managing smart home devices, characterized in that, Including: Collecting device data and fusing it to form a unified set of device status data; Based on the set of device status data, performing feature extraction and combining to generate a multi-dimensional life prediction feature vector; According to the multi-dimensional life prediction feature vector, constructing a device life prediction model and real-time outputting the predicted value of the remaining life of the device; Based on the predicted value of the remaining life of the device, identifying the abnormal decline in the device life and triggering a device maintenance warning message; According to the device maintenance warning message, receiving user feedback and adaptively adjusting the parameters of the device life prediction model; The generating of the multi-dimensional life prediction feature vector includes weighted summing the multi-band vibration envelope energy sequences after data processing according to preset weights to obtain vibration fatigue features; Vibration fatigue characteristics Obtained by wavelet packet decomposition and energy weighting: ； Among them, is the vibration fatigue feature at time moment, is the vibration envelope energy of the th frequency band at time ; is the vibration weight of the th frequency band, representing its diagnostic importance or feature sensitivity; is the frequency band index number; Performing integral processing on the temperature time series curve after data processing to obtain cumulative heat load features; Performing square integral on the speed change rate time series after data processing to obtain speed excitation features; The device cumulative heat load features and speed excitation features are respectively defined as: ； Among them, is the cumulative heat load characteristic within the interval, reflecting the heat load; is the temperature at the time point ; is the rate of change of velocity at the time point , is its squared value, used to evaluate the contribution of mechanical excitation energy, indicating the impact of shock or mutation on life; is the velocity excitation characteristic of the rate of change of velocity within the interval; Accumulating the absolute values of the differences between adjacent sampling points of the current waveform data after data processing to obtain current disturbance features; Calculating the current fluctuation feature within the sampling period: ； Among them, is the current disturbance characteristic within a unit time; is the time at the moment, the current value at the th sampling point; is the sampling point index, ; is the power coefficient of the disturbance intensity, reflecting the sensitivity of the system to the amplitude of current fluctuations; Performing logarithmic mapping and normalization processing on the instantaneous power data after data processing to obtain power response features; Combining the vibration fatigue features, cumulative heat load features, speed excitation features, current disturbance features and power response features in a preset order to form a multi-dimensional life prediction feature vector; The constructing of the device life prediction model includes inputting the multi-dimensional life prediction feature vector into a preset set of life prediction functions, and the set of life prediction functions includes a non-linear mapping function for vibration fatigue response features, an incomplete integral function for cumulative heat load features and speed excitation features, a logarithmic normalization function for power response features, and a radical decay function for adjusting the influence degree of current disturbance features; Performing weighted combination on the output results of the set of life prediction functions and performing integral accumulation to form the health loss value of the device under the current operating state; Combining the state disturbance complexity factor during the device operation, performing normalization processing on the health loss value, and outputting the predicted value of the remaining life of the device representing the remaining life status of the device.

2. The smart home device management method according to claim 1, wherein The collecting of the device data includes respectively deploying data collection units in each device in the smart home environment, and the data collection units include vibration sensors, temperature sensors, acceleration sensors, current sensors and power sensors; The vibration sensor is used to collect vibration signals during the operation of the device and generate a multi-band vibration envelope energy sequence through multi-band envelope analysis; The temperature sensor is used to collect the temperature changes of the key components of the device in real time and form a temperature time series curve; The acceleration sensor is used to measure the acceleration changes of the device and obtain the speed change rate time series through numerical calculation; The current sensor is used to measure the current changes during the operation of the device in real time and generate current waveform data; The power sensor is used to measure the power output of the device in real time and generate instantaneous power data; The data acquisition unit periodically sends the multi-band vibration envelope energy sequence, temperature time-series curve, speed change rate time-series, current waveform data, and instantaneous power data to the smart home local gateway through a local wireless communication device for data storage. The smart home local gateway performs preliminary time-stamping processing on the received data to form raw device operation data with a unified timestamp.

3. The smart home device management method according to claim 2, wherein The fusion to form a unified device status data set includes, based on the raw device operation data stored in the smart home local gateway, using a time-series data interpolation algorithm to eliminate the differences in data sampling frequencies of different devices and unify the data sampling frequency; Standardize the raw device operation data through a data standardization algorithm to eliminate the differences in data dimensions and value ranges, and generate standardized device operation data; Use a multi-source data spatio-temporal alignment algorithm for the standardized device operation data, with a unified device identifier and a unified timestamp as the data matching benchmarks, and integrate the data item by item according to the timestamp to form multi-dimensional device operation data with consistent space and time; Use the Kalman filter algorithm to perform data noise reduction processing on the multi-dimensional device operation data to eliminate the measurement errors of various sensors and environmental noise interference, and generate noise-reduced multi-dimensional device operation data; Fuse the noise-reduced multi-dimensional device operation data according to a preset weight through a multi-dimensional data weighted fusion algorithm to obtain a unified data format and output it as a unified device status data set; The device status data set includes the multi-band vibration envelope energy sequence after data processing, the temperature time-series curve after data processing, the speed change rate time-series after data processing, the current waveform data after data processing, and the instantaneous power data after data processing.

4. The smart home device management method according to claim 3, characterized in that, The identification of abnormal decline in device life includes constructing a dynamic health index baseline for device health level based on the type of the target device and historical operation data; Periodically calculate the residual difference between the predicted value of the remaining life of the current device and the dynamic health index baseline, and use a sliding time window to accumulate the residuals. If the continuous period exceeds the set threshold and the accumulated residual value exceeds the set threshold, it is determined that there is an abnormal decline trend in life; Divide multiple levels of maintenance warning levels according to the degree of residuals, including primary warning, intermediate warning, and emergency warning. The primary warning is used to trigger device self-check and user prompt, the intermediate warning is used to limit the device operation ability and record abnormal energy consumption, and the emergency warning is used to automatically cut off the power supply of high-risk devices and push remote maintenance notifications; After the warning level is confirmed, automatically generate device maintenance warning information, which includes the change trend of the predicted value, the current level classification, and recommended treatment measures.

5. The smart home device management method according to claim 4, wherein The adaptive adjustment of the device life prediction model parameters includes, after the device maintenance warning information is pushed, receiving feedback information provided by the user through the smart home terminal, and the feedback information includes the confirmed maintenance execution situation by the user, the actual operation status of the device, the user's subjective evaluation, and other relevant information; Compare the feedback information with the predicted value of the remaining life of the device. If a deviation is found between the predicted value and the actual device state, an adaptive adjustment algorithm is used to optimize the parameters of the life prediction model.

6. A smart home device management system for implementing the smart home device management method according to any one of claims 1 to 5, characterized in that, It includes: Data acquisition module: Collect device data and fuse it to form a unified set of device state data; Feature extraction module: Based on the set of device state data, perform feature extraction and combine to generate a multi-dimensional life prediction feature vector; Life prediction module: Construct a device life prediction model based on the multi-dimensional life prediction feature vector and output the predicted value of the remaining life of the device in real time; Abnormality recognition and warning module: Based on the predicted value of the remaining life of the device, recognize the abnormal decrease in the device life and trigger a device maintenance warning message; Feedback learning module: Receive user feedback according to the device maintenance warning message and adaptively adjust the parameters of the device life prediction model.

Images