Old people safety early warning method and system based on non-intrusive power consumption behavior analysis

Abstract

This invention discloses a method and system for elderly safety early warning based on non-invasive electricity behavior analysis. The method includes: acquiring electricity consumption data from the smart meter of the residence of an elderly person living alone to obtain initial data; performing in-depth analysis of the initial electricity consumption data to determine the elderly person's circadian rhythm, activity level, and characteristics of commonly used electrical appliances, and establishing a personalized electricity behavior model; acquiring real-time electricity consumption data from the smart meter of the residence of the elderly person living alone; comparing the real-time electricity consumption data with the electricity behavior model to identify high-risk abnormal patterns to determine the degree of abnormality; generating different levels of early warning information based on the degree of abnormality, triggering different intervention measures from family reminders to emergency responses. Implementing this method not only achieves a transformation from simple behavior monitoring to comprehensive health care, but also has advantages such as low cost, high acceptance, and ease of promotion.

Description

Technical Field

[0001] This invention relates to the field of health monitoring technology, and more specifically to a method and system for early warning of elderly safety based on non-invasive electricity behavior analysis. Background Technology

[0002] With the increasing aging of society, the safety of elderly people living alone is receiving growing attention. Traditional safety monitoring methods mainly rely on video surveillance, wearable devices, or emergency call systems. However, these methods have significant limitations: video surveillance infringes on privacy, wearable devices require users to wear them and may be forgotten, while emergency call systems only function when the elderly person is conscious and capable of operating them. Therefore, there is an urgent need for a new method that can effectively monitor the living conditions of elderly people living alone while fully protecting their privacy.

[0003] In recent years, NILM (Non-Intrusive Load Monitoring) technology has matured, offering new solutions to the aforementioned problems. By collecting electricity consumption data from household appliances using smart meters and analyzing this data using machine learning algorithms, it's possible to identify the operating modes of different appliances and the user's activity patterns. Currently, this technology is used in energy management and energy efficiency assessment. However, applying NILM to monitor the living conditions of elderly people living alone is still in the exploratory stage, and a mature solution has not yet been developed. Existing research mainly focuses on detecting specific events, while further development is needed for more complex lifestyle pattern recognition and abnormal situation early warning.

[0004] Therefore, it is necessary to design a new method that can accurately identify abnormal behavior patterns and issue timely warnings without installing cameras or requiring the elderly to wear any devices, while fully protecting personal privacy. This not only realizes the transformation from simple behavior monitoring to comprehensive health care, but also has the advantages of low cost, high acceptance, and easy promotion, and is suitable for application scenarios at multiple levels, including government, communities, and elderly care operators. Summary of the Invention

[0005] The purpose of this invention is to overcome the shortcomings of the prior art and provide a safety early warning method and system for the elderly based on non-invasive electricity behavior analysis.

[0006] To achieve the above objectives, the present invention adopts the following technical solution: a safety early warning method for the elderly based on non-invasive electricity behavior analysis, comprising:

[0007] Obtain electricity consumption data from the smart meters in the residences of elderly people living alone to obtain initial data;

[0008] The initial electricity consumption data were analyzed in depth to determine the elderly person's daily routine, activity level, and characteristics of commonly used electrical appliances, and to establish a personalized electricity consumption behavior model.

[0009] Obtain real-time electricity consumption data from the smart meters in the residences of elderly people living alone;

[0010] Real-time electricity consumption data is compared with the electricity consumption behavior model to identify high-risk abnormal patterns and obtain the degree of abnormality; wherein, the high-risk abnormal patterns include at least one of the following: no activity, disappearance of rhythm, long-term absence from home, and abnormal increase in nighttime activity;

[0011] Different levels of early warning information are generated based on the degree of abnormality, triggering different intervention measures ranging from family reminders to emergency responses.

[0012] The further technical solution is as follows: Deep analysis of the initial electricity consumption data is performed to determine the elderly person's circadian rhythm, activity level, and characteristics of commonly used electrical appliances, establishing a personalized electricity consumption behavior model, including:

[0013] The initial electricity consumption data is analyzed in depth. The wake-up time is determined by analyzing the first continuous electricity consumption activity in the morning, the sleep time is inferred based on the low electricity consumption at night, and the peak use of electrical appliances during mealtimes is identified to understand the daily routine of the elderly.

[0014] Calculate the average power or electricity consumption during a specific time period during the day to assess the intensity of activities, and determine whether daily life is normal by monitoring the stability of the base load;

[0015] Identify the electrical appliances and their characteristics commonly used by the elderly, and construct personalized electricity usage behavior models.

[0016] The further technical solution is as follows: the electrical appliance includes at least one of an electric kettle, a television set, and a rice cooker.

[0017] The further technical solution is as follows: comparing real-time electricity consumption data with the electricity consumption behavior model to identify high-risk abnormal patterns in order to obtain the degree of abnormality, including:

[0018] The real-time electricity consumption data is input into the electricity consumption baseline profile model for analysis, in order to compare the real-time electricity consumption data with the work and rest patterns, activity levels and equipment usage patterns in the electricity consumption baseline profile model, identify deviations, and determine the degree of deviation.

[0019] Detect abnormal patterns in wakefulness, circadian rhythm, outdoor activities, and nighttime activities to determine if there are any risks, and identify abnormal patterns;

[0020] The degree of anomaly is determined based on the anomaly pattern and the degree of deviation.

[0021] The further technical solution is as follows: the abnormal mode includes at least one of the following: no activity abnormal mode, rhythm disappearance abnormal mode, long-term absence abnormal mode, and abnormal increase in nighttime activity abnormal mode. The no activity abnormal mode refers to the absence of the expected peak electricity consumption after waking up; the rhythm disappearance abnormal mode refers to the lack of normal life rhythm characteristics in the electricity consumption curve throughout the day; the long-term absence abnormal mode refers to the electricity load remaining at the baseline level for a long time, showing no signs of human activity; and the abnormal increase in nighttime activity abnormal mode refers to the frequency or duration of nighttime activities being higher than usual.

[0022] The further technical solution is as follows: Based on the degree of abnormality, different levels of early warning information are generated, triggering different intervention measures ranging from family reminders to emergency responses, including:

[0023] The warning level is automatically determined based on the degree of abnormality.

[0024] Based on the aforementioned warning level, targeted warning information is generated, triggering different intervention measures ranging from family reminders to emergency responses.

[0025] The further technical solution is as follows: Based on the warning level, targeted warning information is generated, triggering different intervention measures ranging from family reminders to emergency responses, including:

[0026] Based on the warning level, targeted warning information is generated. When the warning level is low, a gentle reminder is sent to community volunteers or children through the application, suggesting that they pay attention to the elderly's living conditions. When the warning level is medium to high, a work order is created and the community grid worker or elderly care consultant is notified to contact the elderly by phone or visit them at home, and feedback on the handling is requested. When the warning level is an emergency alarm level, other sensors are linked to notify emergency contacts and the community emergency service department.

[0027] This invention also provides a safety early warning method for the elderly based on non-invasive electricity behavior analysis, including:

[0028] The first acquisition unit is used to acquire relevant electricity consumption data of the smart meter in the residence of the elderly living alone in order to obtain initial data;

[0029] A unit is established to perform in-depth analysis of the initial electricity consumption data, determine the elderly person's daily routine, activity level, and characteristics of commonly used electrical appliances, and establish a personalized electricity consumption behavior model.

[0030] The second acquisition unit is used to acquire real-time electricity consumption data of the smart meter in the residence of the elderly living alone.

[0031] An anomaly identification unit is used to compare real-time electricity consumption data with the electricity consumption behavior model to identify high-risk anomaly patterns in order to obtain the degree of anomaly; wherein, the high-risk anomaly pattern includes at least one of no activity, disappearance of rhythm, prolonged absence from home, and abnormal increase in nighttime activity;

[0032] The early warning unit is used to generate different levels of early warning information based on the degree of abnormality, triggering different intervention measures ranging from family reminders to emergency responses.

[0033] The further technical solution is as follows: the establishment unit includes:

[0034] The analysis subunit is used to perform in-depth analysis of the initial electricity consumption data, determine the wake-up time by analyzing the first continuous electricity consumption activity in the morning, infer the fall-up time based on the low electricity consumption at night, and identify the peak use of electrical appliances during mealtimes to understand the daily routine of the elderly.

[0035] The calculation subunit is used to calculate the average power or electricity consumption during a specific time period of the day to assess the intensity of activities, and to determine whether daily life is normal by monitoring the stability of the base load.

[0036] The feature recognition subunit is used to identify the electrical appliances and their characteristics that elderly people frequently use, and to build a personalized fingerprint of their electricity usage behavior.

[0037] The further technical solution is as follows: the anomaly identification unit includes:

[0038] The deviation identification subunit is used to input the real-time electricity consumption data into the electricity consumption baseline profile model for analysis, so as to compare the real-time electricity consumption data with the work and rest patterns, activity levels and equipment usage patterns in the electricity consumption baseline profile model, identify deviations, and determine the degree of deviation.

[0039] The pattern detection subunit is used to detect abnormal patterns of wake-up, circadian rhythm, going out, and nighttime activities to determine whether there is a risk and to identify abnormal patterns.

[0040] The degree determination subunit is used to determine the degree of anomaly based on the anomaly pattern and the degree of deviation.

[0041] The advantages of this invention compared to existing technologies are as follows: This invention analyzes the electricity consumption data of smart meters in the homes of elderly people living alone in a non-invasive manner, without requiring the installation of cameras or the elderly to wear any devices. This allows for the creation of personalized electricity consumption behavior models while fully protecting personal privacy. By comparing and analyzing real-time electricity consumption data with this model, high-risk abnormal patterns can be identified, including inactivity, loss of circadian rhythm, prolonged absences from home, and abnormally increased nighttime activity. Different levels of warning information are generated based on the severity of the abnormality, triggering various intervention measures ranging from family reminders to emergency responses. This method not only achieves a transformation from simple behavioral monitoring to comprehensive health care, but also, due to its low cost, high acceptance, and ease of promotion, is applicable to multiple application scenarios at the government, community, and elderly care operators levels, providing effective technical support and social value.

[0042] The present invention will be further described below with reference to the accompanying drawings and specific embodiments. Attached Figure Description

[0043] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the following description of the embodiments will be briefly introduced. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0044] Figure 1 This is a flowchart illustrating the elderly safety early warning method based on non-invasive electricity behavior analysis provided in an embodiment of the present invention.

[0045] Figure 2 A schematic block diagram of an elderly safety early warning system based on non-invasive electricity behavior analysis provided in an embodiment of the present invention;

[0046] Figure 3 A schematic block diagram of a computer device provided for an embodiment of the present invention. Detailed Implementation

[0047] 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.

[0048] It should be understood that, when used in this specification and the appended claims, the terms "comprising" and "including" indicate the presence of the described features, integrals, steps, operations, elements and / or components, but do not exclude the presence or addition of one or more other features, integrals, steps, operations, elements, components and / or collections thereof.

[0049] It should also be understood that the terminology used in this specification is for the purpose of describing particular embodiments only and is not intended to limit the invention. As used in this specification and the appended claims, the singular forms “a,” “an,” and “the” are intended to include the plural forms unless the context clearly indicates otherwise.

[0050] It should also be further understood that the term "and / or" as used in this specification and the appended claims refers to any combination of one or more of the associated listed items and all possible combinations, and includes such combinations.

[0051] Please see Figure 1 , Figure 1 This is a flowchart illustrating a non-invasive electricity behavior analysis-based safety early warning method for the elderly, provided in an embodiment of the present invention. This method is applied to a server. It acquires electricity usage data from the living environment of elderly people living alone using smart meters and employs deep learning algorithms to establish personalized electricity behavior models. This method eliminates the need for cameras or specific devices worn by the elderly, accurately identifying various high-risk abnormal patterns, including inactivity, loss of circadian rhythm, prolonged absences, and abnormal nighttime activity, while protecting privacy. It automatically triggers different levels of early warning information based on the severity of the abnormality, ranging from mild reminders to emergency responses, thus transforming simple behavior monitoring into comprehensive health care. This method is not only low-cost, easy to accept and promote, but also applicable to various scenarios including government, communities, and elderly care operators, providing an efficient and privacy-respecting solution to ensure the safety and health of the elderly.

[0052] Figure 1 This is a flowchart illustrating the elderly safety early warning method based on non-invasive electricity behavior analysis provided in an embodiment of the present invention. Figure 1 As shown, the method includes the following steps S110 to S150.

[0053] S110. Obtain relevant electricity consumption data from the smart meter at the residence of an elderly person living alone to obtain initial data.

[0054] In this embodiment, the initial data refers to high-frequency electricity consumption data collected over a period of time (usually a 2-4 week learning period) from the smart meters installed in the residence of the elderly person living alone. This data includes, but is not limited to:

[0055] Active power: The effective power used by electrical appliances is recorded at regular intervals (e.g., once every 15 minutes), reflecting the power consumption in different time periods.

[0056] Electricity consumption sequence: The total electricity consumption is recorded at fixed time intervals to form a continuous electricity load curve, which is used to analyze the daily electricity consumption habits and patterns of the elderly.

[0057] Continuous power load curve: Based on the above active power and power consumption data, a curve is generated reflecting the household's power usage throughout the day and at all times. This curve not only reveals the basic daily routines of the elderly (such as waking up, eating, and going to bed), but also reflects their usage habits of specific electrical appliances.

[0058] These initial data form the basis for building personalized "electricity consumption behavior fingerprints." In-depth analysis of this data allows for the identification of an elderly person's circadian rhythms, activity levels, and characteristics of frequently used appliances. For example, the first instance of sustained, stable electricity consumption in the morning may indicate the age of wake-up; while the lowest electricity consumption at night may represent sleep duration. Furthermore, identifying specific appliances (such as an electric kettle used at high power for short periods or a television used during specific evening hours) helps to more accurately understand the elderly person's daily activity patterns, thereby establishing a personalized electricity consumption behavior model. This model will be used in subsequent steps for real-time monitoring and anomaly pattern recognition, ensuring timely detection and early warning of potential risks.

[0059] S120. Perform in-depth analysis on the initial electricity consumption data to determine the elderly person's daily routine, activity level, and characteristics of commonly used electrical appliances, and establish a personalized electricity consumption behavior model.

[0060] In this embodiment, the electricity consumption behavior model refers to a personalized model that reflects the daily routines and habits of an elderly person living alone, constructed through in-depth analysis of the initial electricity consumption data collected from the smart meter in their home. This model not only includes the elderly person's basic circadian rhythm (such as wake-up, meal, and sleep times), but also identifies their daily activity levels and characteristics of commonly used appliances, thus providing a basis for subsequent anomaly detection.

[0061] In one embodiment, step S120 described above may include steps S121 to S123.

[0062] S121. Perform in-depth analysis on the initial electricity consumption data, determine the wake-up time by analyzing the first continuous electricity consumption activity in the morning, infer the fall-off time based on the low electricity consumption at night, and identify the peak use of electrical appliances during mealtimes to understand the daily routine of the elderly.

[0063] In this embodiment, in-depth analysis of initial electricity consumption data is performed to determine the elderly person's lifestyle and habits:

[0064] Wake-up time: Determined by analyzing the first sustained and consistent electrical activity in the morning. For example, the use of boiling water or lighting equipment may mark the start of the day.

[0065] Sleep time: This is inferred from the period when the nighttime electricity load is at its lowest level, which usually means that there is no other activity besides basic equipment such as refrigerators and routers.

[0066] Mealtimes: Identify the correlation between peak electricity consumption and specific time periods to understand habitual electricity usage patterns during breakfast, lunch, and dinner. For example, the usage of kitchen appliances such as rice cookers and microwave ovens.

[0067] S122. Calculate the average power or electricity consumption during a specific time period during the day to assess the intensity of activities, and determine whether daily life is normal by monitoring the stability of the base load.

[0068] In this embodiment, the average power or electricity consumption during a specific time period of the day is calculated to assess the daytime activity level of the elderly and monitor the stability of the baseline load.

[0069] Daytime activity level: Select the average power or total electricity consumption during a typical time period (e.g., 8:00 to 20:00) as an indicator to reflect the intensity of the elderly’s daytime activities.

[0070] Baseline load stability: Whether the baseline formed by long-term monitoring of appliances such as refrigerators that are turned on all year round remains stable is an important basis for judging whether daily life is normal.

[0071] S123. Identify the electrical appliances commonly used by the elderly and their characteristics, and construct a personalized electricity usage behavior model.

[0072] In this embodiment, the electrical appliance includes at least one of an electric kettle, a television, and a rice cooker.

[0073] Identify the electrical appliances frequently used by the elderly and their usage patterns, and establish personalized electricity consumption behavior models:

[0074] By analyzing the unique power consumption patterns of different appliances, such as electric kettles (short-term high power), televisions (medium continuous power during specific evening periods), and rice cookers (timed start before meals), a detailed personalized power consumption behavior model can be constructed.

[0075] This personalized model can not only capture the frequency and timing of appliance use, but also understand the underlying lifestyle habits behind these uses, providing important reference for the subsequent identification of abnormal patterns.

[0076] Through the above steps, the system can establish a highly personalized and accurate "electricity consumption behavior fingerprint," thereby enabling automatic and unobtrusive monitoring of the living conditions of elderly people living alone. This non-invasive method effectively protects the privacy of the elderly while promptly detecting and warning of potential risks, reflecting the value of technology for good.

[0077] In this embodiment, a learning period, typically 2 to 4 weeks, is required when establishing the electricity consumption behavior model. During this period, the system conducts in-depth analysis of the initial electricity consumption data collected by the smart meter in the home of an elderly person living alone to construct a normalized model of the elderly person's daily life pattern. This model forms the basis for subsequent anomaly detection and accurately reflects the elderly person's living habits and daily behavioral patterns.

[0078] Extracting circadian rhythms is the first step in building a routine model. By analyzing the first sustained and stable electrical activity in the morning, such as the use of water-boiling or lighting equipment, we determine the elderly person's wake-up time. This time period marks the start of the day and is one of the key indicators for understanding the elderly person's lifestyle. Simultaneously, the system identifies regular peak electricity consumption during breakfast, lunch, and dinner. These peaks are associated with the use of specific appliances, such as the start-up of rice cookers or microwave ovens, helping us understand the elderly person's eating habits and schedules. Furthermore, sleep duration is inferred by observing the start time when nighttime electricity load consistently remains at its lowest level (mainly maintained by basic appliances such as refrigerators and routers). This helps in understanding the elderly person's rest patterns and biological clock.

[0079] Activity level quantification involves calculating average power consumption or total electricity usage during the day (e.g., from 8:00 AM to 8:00 PM) to assess the elderly person's daytime activity level. This quantification method reflects the intensity of their daytime activity and their daily routine. In addition to daytime activity levels, monitoring the stability of baseline load is also crucial. By tracking the stability of the baseline electricity consumption formed by appliances that are frequently used, such as refrigerators, it's possible to determine whether the elderly person's daily life is normal. Any deviation from this routine may be an early sign of potential problems, such as deteriorating health or significant changes in lifestyle habits.

[0080] Finally, habitual device recognition involves identifying specific appliances frequently used by the elderly and their characteristics. Different appliances exhibit different power consumption patterns; for example, electric kettles may be used at high power for short periods, televisions at medium power for extended periods in the evening, and rice cookers may be scheduled to start before meals. By analyzing these appliance usage patterns, the system can construct a detailed, personalized electricity consumption behavior model. This model not only captures the frequency and timing of appliance use but also reveals the underlying lifestyle habits behind these uses, providing important references for further anomaly pattern recognition.

[0081] In summary, through these three steps, the system can establish a highly personalized and accurate "electricity consumption behavior fingerprint," or electricity consumption behavior model. This enables automatic and seamless monitoring of the living conditions of elderly people living alone. This method not only effectively protects the privacy of the elderly but also promptly detects and warns of potential risks without interfering with their daily lives, embodying the value of technology for good.

[0082] S130: Obtain real-time electricity consumption data from the smart meter at the residence of an elderly person living alone.

[0083] This step involves acquiring real-time electricity usage data from the smart meters already installed in the homes of elderly people living alone. This data is typically collected at a high frequency (e.g., every 15 minutes) and includes information such as active power and electricity consumption, forming a continuous electricity load curve. This step ensures that the system can continuously monitor the elderly person's electricity usage behavior, providing data support for subsequent analysis.

[0084] S140. Compare the real-time electricity consumption data with the electricity consumption behavior model to identify high-risk abnormal patterns and obtain the degree of abnormality; wherein, the high-risk abnormal patterns include at least one of no activity, disappearance of rhythm, long-term absence from home and abnormal increase in nighttime activity.

[0085] In this embodiment, the degree of abnormality refers to the risk level quantitatively assessed based on the degree of deviation from normal daily routines, activity levels, and device usage patterns, after comparing real-time electricity consumption data with a personalized electricity consumption baseline profile model. In short, it is an indicator of the severity of how a senior living alone deviates from their daily behavioral patterns.

[0086] In one embodiment, step S140 described above may include steps S141 to S143.

[0087] S141. Input the real-time electricity consumption data into the electricity consumption baseline profile model for analysis, so as to compare the real-time electricity consumption data with the work and rest patterns, activity levels and equipment usage patterns in the electricity consumption baseline profile model, identify deviations, and determine the degree of deviation.

[0088] In this embodiment, the degree of deviation refers to the magnitude of the difference between the real-time monitored electricity consumption data and the pre-established personalized electricity consumption baseline profile, used to quantify the gap between the daily activity patterns of elderly people living alone and normal behavioral benchmarks. In short, it measures the degree of change in actual electricity consumption relative to the expected normal electricity consumption pattern.

[0089] The first step involves inputting the collected real-time electricity consumption data into the previously established personalized electricity consumption baseline profile model for analysis. This process involves comparing the real-time data with the baseline model's patterns of daily routines, activity levels, and device usage to identify any discrepancies and their extent. The goal of this stage is to quantify the difference between actual electricity consumption behavior and typical behavior, providing a basis for subsequent risk assessment.

[0090] S142. Detect abnormal patterns of wakefulness, rhythm, going out, and nighttime activities to determine if there is any risk, and obtain abnormal patterns.

[0091] In this embodiment, the abnormal mode includes at least one of the following: no activity abnormal mode, rhythm disappearance abnormal mode, long-term absence abnormal mode, and abnormally increased nighttime activity abnormal mode. The no activity abnormal mode refers to the absence of the expected peak electricity consumption after waking up; the rhythm disappearance abnormal mode refers to the lack of normal daily rhythm characteristics in the electricity consumption curve throughout the day; the long-term absence abnormal mode refers to the electricity load remaining at the baseline level for a long period of time, showing no signs of human activity; and the abnormally increased nighttime activity abnormal mode refers to the frequency or duration of nighttime activities being higher than usual.

[0092] In this step, the system specifically detects four types of abnormal patterns: abnormal inactivity, abnormal loss of rhythm, abnormal prolonged absence from home, and abnormally increased nighttime activity. Each abnormal pattern has its own specific rules and definitions:

[0093] No activity anomaly mode: refers to the failure to observe the corresponding peak power consumption after the expected wake-up time.

[0094] Abnormal rhythm loss pattern: refers to the lack of normal daily life rhythm characteristics in the electricity consumption curve throughout the day.

[0095] Abnormal pattern of prolonged absence from home: This indicates that the electricity load has remained at the baseline level for an extended period, showing no signs of human activity.

[0096] An abnormal increase in nighttime activity: This means that the frequency or duration of nighttime activity is significantly higher than usual.

[0097] S143. Determine the degree of abnormality based on the abnormal pattern and the degree of deviation.

[0098] Finally, after identifying the specific anomaly pattern, the system further assesses the severity of the anomaly based on the degree of deviation. This step is crucial for triggering appropriate response measures. For example, if a "no activity" anomaly is detected and the deviation is very high, an emergency alarm may be triggered immediately, notifying relevant personnel to take action; while for some milder anomalies, such as a slight increase in nighttime activity, the system may generate a low-risk alert, suggesting that family members or community volunteers pay appropriate attention.

[0099] Through the above steps, the system can not only automatically and seamlessly identify abnormal fluctuations in the living conditions of elderly people living alone, but also provide timely warnings of potential risks, offering crucial support to communities, relatives, and elderly care service institutions, thus embodying the value of technology for good.

[0100] In this embodiment, step S140 compares and analyzes the real-time collected electricity consumption data with a personalized baseline to identify four high-risk abnormal patterns and issue corresponding warnings. First, the "no activity" abnormality refers to the failure to detect a characteristic "wake-up activity" electricity consumption peak within 2 to 3 hours after the elderly person's usual wake-up time. This situation may indicate that the elderly person is experiencing an emergency such as a sudden illness, fall, or coma, and is therefore classified as a high-alert level, requiring immediate telephone contact or home visit verification. Second, the "disappearance of rhythm" abnormality is characterized by an abnormally flat electricity consumption curve throughout the day, lacking obvious peaks for meals and activities, and with total electricity consumption significantly lower than the personalized baseline. This indicates that the elderly person may be bedridden all day, extremely unwell, or has left home, also falling under the high-alert level.

[0101] Furthermore, the "prolonged absence without returning home" anomaly refers to an electrical load that remains at the minimum level for more than 24 hours without any signs of human activity. This could be due to the elderly person going out to visit relatives or seek medical treatment, but there is also the possibility of an accident preventing them from returning home. Therefore, the warning level is medium to high, and it is recommended to contact them promptly to confirm their safety. Finally, the "unusually increased nighttime activity" anomaly is manifested in an abnormally increased frequency of getting up to turn on lights at night, using the bathroom, or exhibiting unusually long periods of electrical usage. This type of anomaly may indicate that the elderly person is suffering from health problems such as insomnia, gastrointestinal issues, or breathing difficulties, suggesting the need to pay attention to their health status. The warning level is medium. By effectively identifying and issuing warnings for these abnormal patterns, potential risks can be detected and addressed in a timely manner, ensuring the safety of elderly people living alone.

[0102] S150. Generate different levels of early warning information based on the degree of abnormality, triggering different intervention measures from family reminders to emergency responses.

[0103] In one embodiment, step S150 described above may include steps S151 to S152.

[0104] S151. Automatically determine the warning level based on the degree of abnormality.

[0105] This step involves real-time analysis of data collected by smart meters and comparison of this data with each elderly person living alone's personalized electricity consumption baseline. Through this comparison, the system can identify four main high-risk abnormal patterns: "inactivity" abnormality, "absence of circadian rhythm" abnormality, "excessive absence from home" abnormality, and "abnormally increased nighttime activity." Each abnormal pattern has its specific risk assessment and corresponding warning level (low, medium-high, high). For example, "inactivity" abnormality is classified as a high warning level because it may mean that the elderly person has encountered an emergency such as a sudden illness or a fall; while "abnormally increased nighttime activity" is classified as a medium warning level, indicating that attention should be paid to the elderly person's health condition.

[0106] S152. Based on the aforementioned warning level, generate targeted warning information and trigger different intervention measures, ranging from family reminders to emergency responses.

[0107] In this embodiment, targeted warning information is generated based on the warning level. When the warning level is low, a gentle reminder is sent to community volunteers or children through the application, suggesting that they pay attention to the elderly person's living conditions. When the warning level is medium to high, a work order is created and the community grid worker or elderly care consultant is notified to contact the elderly person by phone or visit them at home, and feedback on the handling is requested. When the warning level is an emergency alarm level, emergency contacts and community emergency service departments are notified through linkage with other sensors.

[0108] In this step, the system will generate corresponding warning information and take different intervention measures based on the warning level determined in the previous step:

[0109] Low-risk alert: When the alert level is low, the system sends a gentle reminder to community volunteers or the elderly person's children through the application, suggesting that they pay attention to the elderly person's living conditions, but no immediate action is required.

[0110] Medium-to-high risk warning: If the warning level rises to medium-to-high, the system will create a work order and notify the community grid worker or elderly care consultant, requiring them to contact the elderly person by phone or visit them at home to check on their condition and report the results. This ensures that any potential problems receive timely attention and resolution.

[0111] Emergency Alarm: For the highest level of alert, the system will not only notify emergency contacts, but also link other sensors (such as door magnetic sensors) to directly contact the community emergency service department to respond to possible emergencies as quickly as possible, such as an elderly person falling or becoming unconscious.

[0112] Through this tiered early warning mechanism, the model effectively transitions from simple data analysis to practical care intervention, not only improving the safety and security of elderly people living alone but also providing a powerful support tool for communities, family members, and elderly care service institutions. At the same time, this approach protects the privacy of the elderly while ensuring a rapid response to potential risks, fully embodying the concept of "technology for good."

[0113] In this embodiment, by analyzing real-time electricity consumption data from smart meters, abnormal fluctuations in the living conditions of elderly people living alone are automatically identified, triggering a multi-level response mechanism to provide different intervention measures, from family reminders to emergency responses. Low-risk alerts are pushed to the applications of community volunteers or children, reminding them to pay attention to changes in the elderly person's daily routine; medium- and high-risk alerts automatically generate work orders and notify community grid workers or elderly care consultants to contact them by phone or visit them at home, ensuring timely verification of the situation and feedback of the processing results in the system; in the most urgent cases, by combining the "no activity" anomaly with the door magnetic sensor (optional), emergency contacts and community emergency response are directly linked to quickly respond to possible emergencies.

[0114] The innovation and advantages of this method lie in its absolute non-invasiveness and high acceptability. It does not rely on video / audio monitoring, thus perfectly protecting the privacy and dignity of the elderly, making the technology easily accepted and promoted. Furthermore, this method achieves a paradigm shift from "behavior" to "care," upgrading electricity data analysis, originally used to monitor occupancy, into a tool for safeguarding the health and well-being of the elderly, making the technology more humane and socially valuable. Personalization and adaptability allow the model to be built based on each elderly person's historical data, avoiding false alarms caused by uniform thresholds and improving the accuracy of early warnings. Simultaneously, 24 / 7 automatic monitoring can be achieved using existing or low-cost smart meters, greatly alleviating the pressure of manual inspections, reducing costs, and making it potential for large-scale application. More importantly, by forming a "data-response" closed loop, it not only identifies potential problems but also embeds them into community service processes, ensuring that early warnings are effectively handled.

[0115] In terms of application scenarios and business models, this model is suitable for governments and communities as part of smart elderly care solutions, enhancing grassroots elderly care services and improving the level of refined governance. For children and relatives, a paid subscription provides remote parental care, alleviating their worries. For elderly care operators, this model can be integrated into their smart elderly care platforms, reducing care costs while increasing service value. For power or property management companies, this service not only enhances customer loyalty but also fulfills social responsibility, reflecting the company's social value. In summary, the method in this embodiment transforms seemingly cold, impersonal electricity data into a "digital beacon" protecting the safety of elderly people living alone, keenly sensing their life rhythms. This is not only a technological innovation but also a model of technology for good, demonstrating enormous social significance and market potential.

[0116] In one embodiment, the following steps may be added before step S110:

[0117] Acquire multi-source sensing data of the residence of elderly people living alone, including electricity consumption data from smart meters, non-contact radar monitoring data, and smart access control status data;

[0118] The smart meter electricity consumption data, the non-contact radar monitoring data, and the smart access control status data are fused according to the preset data credibility weight to obtain fused home activity data, which is used as the initial data.

[0119] The data credibility weight is dynamically adjusted based on the stability and accuracy of each data source: when the electricity meter data is continuous and stable, its weight is increased; when the radar detects obvious human activity but the electricity meter data is abnormal, the electricity meter weight is decreased and the radar weight is increased; when the access control status conflicts with the electricity consumption data, a manual verification mechanism is introduced.

[0120] In this embodiment, during the stage of acquiring multi-source sensing data of the residence of an elderly person living alone, the system first establishes a three-layer data acquisition architecture to achieve comprehensive and complementary home activity monitoring. The first layer is the smart meter electricity consumption data acquisition layer. By connecting to the smart meter in the residence, electrical parameters such as total power, circuit current, voltage, and power factor are collected at a frequency of minutes or higher, forming continuous electricity consumption time series data. This data can reflect the regularity of the elderly person's use of electrical appliances and the characteristics of their daily rhythm, but it has limitations such as being unable to distinguish between a person's static state and their state of being away from home, and having difficulty capturing low-power activities (such as sitting and reading). The second layer is a non-contact radar monitoring data acquisition layer. Millimeter-wave radar or ultra-wideband radar devices are deployed in key areas of the residence (such as the living room, bedroom, and bathroom entrance). These devices emit electromagnetic waves and receive echo signals, enabling them to penetrate non-metallic obstacles and achieve imperceptible detection of human targets. Specifically, the radar monitoring data includes static human presence signals (detecting micro-movement features such as breathing and heartbeat through the micro-Doppler effect), dynamic motion trajectory signals (calculating human position movement through distance-Doppler-angle three-dimensional information), and human posture classification signals (identifying postures such as standing, sitting, and lying down through point cloud features). The advantage of this data source is that it can achieve presence confirmation and activity intensity assessment without relying on the elderly wearing any devices. However, false alarms or missed detections may occur in complex electromagnetic environments or when multipath effects are present. The third layer is the smart access control status data collection layer. By installing a door magnetic sensor or smart door lock on the entrance door, information such as the door's opening and closing status, opening and closing timestamps, and unlocking methods (such as key, password, fingerprint) are collected in real time. This data provides direct evidence for determining whether the elderly have gone out or returned, but it cannot reflect the activity status inside the door and there are special cases such as the elderly not closing the door properly or leaving without locking the door temporarily.

[0121] After acquiring the three types of raw data mentioned above, the system enters the multi-source data fusion processing stage, the core of which lies in constructing an adaptive data credibility weight allocation mechanism. This mechanism first assigns a basic credibility weight to each data source, typically based on historical data statistics: smart meter data, due to its stable acquisition and accurate quantification, is given a higher basic weight (e.g., 0.5); non-contact radar data, because it can directly sense the presence of humans, is given a medium basic weight (e.g., 0.3); and smart access control data, due to its strong dispersion but clear events, is given a lower basic weight (e.g., 0.2). Based on this, the system introduces a dynamic weight adjustment strategy, achieving adaptive optimization of the weights by monitoring the quality indicators and consistency relationships of each data source in real time.

[0122] Specifically, dynamic weight adjustment includes processing logic for three typical scenarios: The first scenario is the continuous stability confirmation scenario of electricity meter data. When the system detects that the smart meter has complete data sampling, no communication interruption, and power fluctuations that conform to the normal operating characteristics of electrical appliances (i.e., no sudden noise or constant zero value anomalies) within multiple consecutive time windows (such as the past 30 minutes), it determines that the electricity meter data is in a highly reliable state. At this time, its weight coefficient is gradually increased (up to 0.6), while the weights of other data sources are correspondingly reduced, so that the fusion result mainly relies on electricity consumption behavior analysis. The second scenario is the radar-electricity meter data conflict scenario. When the radar monitoring equipment detects obvious human micro-movement characteristics (such as the Doppler frequency shift corresponding to chest cavity fluctuations caused by breathing) or continuous movement trajectories in a specific area, indicating that there is indeed human activity in the area, but the smart meter data at the same time shows extremely low power or zero power (i.e., no electrical appliances are used), the system determines that the electricity meter data may be due to line faults, communication interruptions, or the elderly engaging in extremely low-power activities (such as static activities). If the data is distorted due to sitting or resting, the weight of the electricity meter will be immediately reduced (down to a minimum of 0.2) and the weight of the radar will be increased (up to a maximum of 0.5). This will ensure that the fusion results prioritize the information about the presence of a person perceived by the radar, avoiding missed activity detection due to the failure of a single data source. The third scenario is a conflict between the access control and electricity meter status. When the smart access control record shows that the door is open or has just been unlocked (indicating that the elderly person may have gone out or returned), but the smart meter data still shows that high-power appliances are running indoors (such as air conditioners or TVs that are not turned off), the system will determine that there is a data inconsistency. Possible situations include the elderly person leaving without turning off appliances, visitors being indoors, or false alarms from the access control sensor. Since this conflict involves the accuracy of safety judgment, the system will automatically trigger a manual verification mechanism. On the one hand, the automatic warning decision will be temporarily frozen, and on the other hand, a conflict alarm message will be pushed to the preset community grid member or family member terminal to request manual confirmation of the elderly person's actual status. After receiving manual feedback, the automatic fusion processing will be resumed, and the conflict case will be recorded in the log for subsequent weight strategy optimization.

[0123] Based on dynamic weight adjustments, the system performs weighted fusion calculations to generate unified home activity data. For three types of data sources with the same timestamp, the system first performs time alignment and format standardization, encoding the power value of the electricity meter, the presence confidence and motion intensity value of the radar, and the on / off status of the access control system into a unified data structure. Then, it performs weighted summation or rule-based decision fusion based on the real-time weights of each data source. For example, the human presence judgment can be expressed as the fused presence index = w1 × electricity meter activity index + w2 × radar presence confidence + The w3×access control indoor inference index (where the access control indoor inference index is 1 when the door is closed and there is no record of leaving, 0.5 pending confirmation when the door is open, and 0 when there is a clear record of leaving) is used to generate fused home activity data that represents the elderly’s activity status in the residence in the form of time series. This includes presence labels (indoor / outdoor / uncertain), activity intensity level (still / light activity / moderate activity / vigorous activity), distribution of main activity areas, and circadian rhythm phase (sleep / wake-up / meal / leisure / going out). This fused data serves as the initial input for the subsequent establishment of personalized electricity behavior models and anomaly detection, effectively overcoming the limitations of a single data source and significantly improving the system’s perception robustness and judgment accuracy in complex home environments.

[0124] In another embodiment, after step S120, the method further includes:

[0125] The electricity consumption behavior model is used to extract electricity consumption behavior information types from the circadian rhythm data, activity level data, and common appliance characteristic data.

[0126] Heterogeneous matrix architecture analysis is performed based on the electricity consumption behavior information type to construct a heterogeneous matrix architecture for electricity consumption behavior. The architecture includes: basic matrix nodes with time as the horizontal axis and appliance type as the vertical axis, and unique identifier nodes for heterogeneous matrices labeled with the elderly's activity status.

[0127] The circadian rhythm data, activity level data, and characteristic data of commonly used electrical appliances are transmitted to the heterogeneous matrix architecture of electricity consumption behavior for data mapping to generate a heterogeneous data matrix of electricity consumption behavior.

[0128] After acquiring real-time electricity consumption data, dynamic feature pattern recognition is performed on the real-time electricity consumption data and the heterogeneous data matrix of electricity consumption behavior: dynamic keyframe analysis is performed on the real-time data to generate keyframe dynamic data, time series analysis is performed on the keyframe dynamic data to generate time series keyframe data, and nonlinear dynamic spatiotemporal feature analysis of electricity consumption is performed based on the time series keyframe data to generate dynamic spatiotemporal feature data of electricity consumption.

[0129] Based on the aforementioned dynamic spatiotemporal characteristic data of electricity consumption, abnormal electricity consumption patterns are identified, generating dynamic characteristic pattern data of electricity consumption, which is used to identify the high-risk abnormal patterns.

[0130] In this embodiment, during the electricity consumption behavior information type extraction stage, the system first performs semantic parsing and feature classification on the three core data types in the electricity consumption behavior model to generate structured electricity consumption behavior information types. For circadian rhythm data, the system extracts time phase types (including sleep period, morning wake-up period, breakfast period, morning activity period, lunch period, afternoon nap period, afternoon activity period, dinner period, evening leisure period, nighttime wake-up period, etc.), rhythm stability indicators (standard deviation of start and end times of each time phase, coefficient of variation of duration), and rhythm offset patterns (difference between weekdays and weekends, seasonal adjustment factor), forming a set of behavioral labels in the time dimension. For activity level data, the system extracts power consumption level types (classified as basic standby level 0-50W, low-power activity level 50-200W, moderate activity level 200-800W, and high-power activity level above 800W), activity duration types (short-term pulsed use, medium-term continuous use, and long-term continuous use), and activity frequency types (high-frequency daily use, medium-frequency regular use, low-frequency occasional use, and extremely low-frequency rare use), forming a multi-dimensional description of power consumption intensity and activity characteristics. For common appliance characteristic data, the system extracts appliance identity types (identifying specific appliances such as electric kettles, rice cookers, televisions, air conditioners, and microwave ovens through load imprints), appliance function types (cooking, entertainment, environmental control, lighting, and personal care), and appliance usage association types (co-occurrence patterns such as breakfast combinations of rice cookers and electric kettles, and evening combinations of televisions and lights), forming device-level behavioral semantics. Combining the extraction results of the above three types of data, the system generates a standardized electricity consumption behavior information type library, with each information type assigned a unique code and attribute description, serving as the basic element for subsequent matrix architecture construction.

[0131] In the heterogeneous matrix architecture analysis phase, the system designs a two-layer matrix structure based on electricity consumption behavior information types to achieve unified organization of multi-dimensional data. The first layer is the basic matrix node layer, which constructs a two-dimensional grid architecture with time as the horizontal axis and appliance type as the vertical axis: the horizontal axis discretizes the 24-hour cycle into fixed time slots (e.g., one slot every 15 minutes, for a total of 96 time units), and each time slot is associated with the activity phase label and expected activity level of that period; the vertical axis lists all commonly used appliances identified in the elderly person's residence and their respective functional categories, and each appliance node contains static attributes such as its rated power, typical load curve, and usage frequency statistics. The grid intersections form basic matrix nodes, and each node stores statistical characteristics of the appliance in that period, such as historical usage probability, average usage duration, and power fluctuation range. For example, the "rice cooker - breakfast period" node records information such as the probability of using a rice cooker during breakfast period as 0.85, the average start time as 6:30, and the average power as 800W. The second layer is a heterogeneous matrix unique identifier node layer. This layer establishes a cross-dimensional association index based on the elderly's activity status as a label. The activity status labels include scene categories with clear semantics, such as "deep sleep," "morning preparation," "cooking and dining," "home leisure," "cleaning and tidying," and "leaving home." Each unique identifier node is associated with multiple node combinations in the base matrix layer through hyperlinks. For example, the "morning preparation" identifier node is associated with base nodes such as "electric kettle - morning period" (high probability), "lighting light - morning period" (high probability), and "television - morning period" (low probability), forming a typical electricity consumption pattern template for this activity status. The system further associates the unique identifier nodes of the heterogeneous matrix with the nodes of the basic matrix, analyzes the conditional dependencies (e.g., the use of a rice cooker is usually accompanied by the use of an electric kettle), mutual exclusions (e.g., cooking appliances are rarely used during deep sleep), and temporal succession relationships (e.g., cooking usually follows preparations in the morning), generates heterogeneous matrix node association data, and establishes a complete heterogeneous matrix architecture for electricity consumption behavior, realizing a structured mapping from discrete appliance usage records to high-level activity semantics.

[0132] During the data mapping phase, the system populates specific data instances from the electricity consumption behavior model into a heterogeneous matrix architecture to generate a heterogeneous data matrix for electricity consumption behavior. For circadian rhythm data, the system calculates the probability of circadian rhythm phase attribution for each time slot based on historical time series, and maps the probability values ​​to the attribute fields of the horizontal axis time slots. For activity level data, the system calculates indicators such as average power, power variance, and load duration within each time slot, and maps them to the power consumption feature fields of the corresponding time slots. For characteristic data of commonly used appliances, the system constructs a usage event sequence for each appliance, marks the start and end time slots of the events, the actual power curve, and the overlap relationship with other appliances, and maps them to the appliance nodes on the vertical axis and their associated base matrix nodes. At the same time, the system assigns instantiation weights to the unique identifier nodes of the heterogeneous matrix based on actual data. For example, if historical data shows that an elderly person uses an electric kettle in the morning for 30 consecutive days, the association weight between the "Morning Preparation" identifier node and the "Electric Kettle - Morning Period" base node is set to a high confidence level of 0.95. The resulting heterogeneous data matrix of electricity consumption behavior is a sparse and high-dimensional tensor structure that not only preserves the time-series characteristics of the original electricity consumption data, but also realizes the semantic association between electrical appliances, time, and activity status through the matrix architecture, providing a structured prior knowledge base for subsequent dynamic pattern recognition.

[0133] In the dynamic feature pattern recognition stage, the system first performs dynamic keyframe analysis on real-time power consumption data to extract key nodes of state transitions. Keyframe extraction is based on the detection of significant changes in power consumption status: when the real-time power curve shows a step increase or decrease exceeding a threshold (e.g., 200W), it is marked as an event keyframe, indicating that an appliance is starting or stopping; when the power fluctuation characteristics undergo a sustained change (e.g., from stable to drastic fluctuation), it is marked as a behavior keyframe, indicating a change in activity type; when multiple appliances start successively within a short time window (e.g., 5 minutes), it is marked as a combination keyframe, indicating the occurrence of a complex activity scenario. Each keyframe records attributes such as timestamp, power value, the set of appliances involved, and continuous state, forming a dynamic data sequence of keyframes. Subsequently, the system performs time-series analysis on the keyframe dynamic data, constructs a local keyframe sequence through a sliding time window (such as the past 2 hours), analyzes the time-series patterns in the sequence (such as whether the keyframe interval is regular, whether the appliance start-up order is fixed, and whether the power peak decays), and generates time-series keyframe data. This data includes both the absolute time position of the keyframe and the relative time-series relationship between keyframes (such as "the rice cooker starts an average of 8 minutes after the electric kettle starts").

[0134] Based on time-series keyframe data, the system further performs nonlinear spatiotemporal characteristic analysis of electricity consumption dynamics. Spatially, the system maps discrete keyframes to an activity state label space through node association relationships in a heterogeneous data matrix, determining which unique identifier node's typical pattern best matches the current keyframe combination. For example, if "electric kettle + rice cooker" are detected to be started sequentially in the morning, it is mapped to the "morning preparation" activity state, achieving a nonlinear transition from lower-level electricity consumption events to higher-level activity semantics. Temporally, the system analyzes the deviation of the keyframe sequence from historical circadian rhythm patterns, calculates the offset of the current moment relative to the expected circadian rhythm phase (e.g., expected 6:30 AM, actual 7:15 AM morning pattern, offset +45 minutes), and assesses whether this offset exceeds the normal variation range. In the spatiotemporal joint dimension, the system constructs dynamic feature trajectories to track the elderly's state transition paths from sleep to waking up, and from home to going out. It analyzes the smoothness (whether there are abnormal transitions, such as jumping directly from sleep to going out without morning preparation), completeness (whether expected intermediate states are missing), and repeatability (whether there are abnormal oscillations). Using kernel methods or deep learning encoders, the system projects these spatiotemporal features into a high-dimensional feature space, capturing the nonlinear dynamic patterns in electricity consumption behavior and generating dynamic spatiotemporal feature data of electricity consumption. This data represents the elderly's current behavioral state and its evolutionary trend in the time-activity space in a compact vector form.

[0135] Finally, the system identifies abnormal electricity consumption patterns based on dynamic spatiotemporal characteristic data, generating dynamic electricity consumption feature pattern data. The identification process employs a multi-level matching strategy: first, at the feature vector level, it calculates the similarity between real-time features and activity state templates in the heterogeneous data matrix to identify the pattern category to which the current behavior belongs; second, at the trajectory level, it compares the compliance of the current state transition path with the normal behavior graph, detecting any illegal transitions or missing nodes; third, at the prediction level, it infers the expected behavior for the next period based on historical dynamic models, assessing the degree of deviation between the actual behavior and the prediction. Based on the above analysis, the system outputs dynamic electricity consumption feature pattern data, including current activity state labels, state confidence, rhythm compliance scores, and abnormal risk indicators, which are directly used to identify high-risk abnormal patterns, such as no-activity anomalies (no keyframe sequence matching any activity state detected for a long time), rhythm disappearance anomalies (dynamic features randomly walk in the state space without clear phase assignment), long-term absence anomalies (no return pattern detected after an absence transfer for a long time), and increased nighttime activity anomalies (frequent activity state transitions that should belong to the daytime during nighttime). Compared to traditional threshold comparison methods, this dynamic feature pattern data can capture complex spatiotemporal correlations and evolutionary anomalies in electricity consumption behavior, significantly improving the early warning capability for safety risks to the elderly.

[0136] The aforementioned method for elderly safety early warning based on non-invasive electricity behavior analysis analyzes the electricity consumption data of smart meters in the homes of elderly people living alone without requiring cameras or any devices worn by the elderly. This allows for the creation of personalized electricity behavior models while fully protecting personal privacy. By comparing real-time electricity consumption data with this model, high-risk abnormal patterns can be identified, including inactivity, loss of circadian rhythm, prolonged absences from home, and abnormally increased nighttime activity. Different levels of early warning information are generated based on the severity of the abnormality, triggering various intervention measures from family reminders to emergency responses. This method not only represents a shift from simple behavioral monitoring to comprehensive health care but also, due to its low cost, high acceptance, and ease of promotion, is applicable to multiple application scenarios at the government, community, and elderly care operators levels, providing effective technical support and social value.

[0137] Figure 2 This is a schematic block diagram of an elderly safety early warning system 300 based on non-invasive electricity behavior analysis provided in an embodiment of the present invention. Figure 2 As shown, corresponding to the above-described method for elderly safety warnings based on non-invasive electricity consumption behavior analysis, this invention also provides an elderly safety warning system 300 based on non-invasive electricity consumption behavior analysis. This elderly safety warning system 300 includes a unit for executing the above-described method for elderly safety warnings based on non-invasive electricity consumption behavior analysis, and the system can be configured in a server. Specifically, please refer to... Figure 2 The elderly safety early warning system 300 based on non-invasive electricity behavior analysis includes a first acquisition unit 301, an establishment unit 302, a second acquisition unit 303, an anomaly identification unit 304, and an early warning unit 305.

[0138] The first acquisition unit 301 is used to acquire relevant electricity consumption data from the smart meter of the residence of the elderly living alone to obtain initial data; the establishment unit 302 is used to perform in-depth analysis on the initial electricity consumption data to determine the elderly's circadian rhythm, activity level, and characteristics of commonly used electrical appliances, and to establish a personalized electricity consumption behavior model; the second acquisition unit 303 is used to acquire real-time electricity consumption data from the smart meter of the residence of the elderly living alone; the anomaly identification unit 304 is used to compare the real-time electricity consumption data with the electricity consumption behavior model to identify high-risk abnormal patterns to obtain the degree of abnormality; wherein, the high-risk abnormal patterns include at least one of no activity, disappearance of rhythm, prolonged absence from home, and abnormally increased nighttime activity; the early warning unit 305 is used to generate different levels of early warning information according to the degree of abnormality, triggering different intervention measures from family reminders to emergency responses.

[0139] In one embodiment, the establishment unit 302 includes:

[0140] The analysis subunit is used to perform in-depth analysis of the initial electricity consumption data, determine the wake-up time by analyzing the first continuous electricity consumption activity in the morning, infer the fall-off time based on the low electricity consumption at night, and identify the peak use of electrical appliances during mealtimes to understand the elderly's daily routine patterns; the calculation subunit is used to calculate the average power or electricity consumption during a specific time period during the day to assess the activity intensity, and determine whether daily life is normal by monitoring the stability of the baseline load; the feature recognition subunit is used to identify the electrical appliances and their characteristics commonly used by the elderly, and construct a personalized electricity consumption behavior fingerprint.

[0141] In one embodiment, the anomaly detection unit 304 includes:

[0142] The deviation identification subunit is used to input the real-time electricity consumption data into the electricity consumption baseline profile model for analysis, so as to compare the real-time electricity consumption data with the work-rest patterns, activity levels and equipment usage patterns in the electricity consumption baseline profile model, identify deviations, and determine the degree of deviation; the pattern detection subunit is used to detect abnormal patterns of wake-up, rhythm, going out and nighttime activities, determine whether there is a risk, and obtain abnormal patterns; the degree determination subunit is used to determine the degree of abnormality based on the abnormal patterns and the degree of deviation.

[0143] In one embodiment, the early warning unit 305 includes:

[0144] The assessment subunit is used to automatically assess the warning level based on the degree of abnormality; the information generation subunit is used to generate targeted warning information based on the warning level, triggering different intervention measures from family reminders to emergency responses.

[0145] In one embodiment, the information generation subunit is used to generate targeted warning information based on the warning level. When the warning level is low, a gentle reminder is sent to community volunteers or children through the application, suggesting that they pay attention to the elderly person's living conditions. When the warning level is medium to high, a work order is created and the community grid worker or elderly care consultant is notified to contact the elderly person by phone or visit them at home, and feedback on the handling is requested. When the warning level is an emergency alarm level, emergency contacts and community emergency service departments are notified through linkage with other sensors.

[0146] It should be noted that those skilled in the art can clearly understand that the specific implementation process of the elderly safety early warning system 300 based on non-intrusive electricity behavior analysis and each unit can be referred to the corresponding description in the foregoing method embodiments. For the sake of convenience and brevity, it will not be repeated here.

[0147] The aforementioned elderly safety early warning system 300 based on non-invasive electricity behavior analysis can be implemented as a computer program, which can be used in, for example... Figure 3 It runs on the computer device shown.

[0148] Please see Figure 3 , Figure 3 This is a schematic block diagram of a computer device provided in an embodiment of this application. The computer device 500 can be a server, wherein the server can be a standalone server or a server cluster composed of multiple servers.

[0149] See Figure 3 The computer device 500 includes a processor 502, a memory, and a network interface 505 connected via a system bus 501. The memory may include a non-volatile storage medium 503 and internal memory 504.

[0150] The non-volatile storage medium 503 may store an operating system 5031 and a computer program 5032. The computer program 5032 includes program instructions that, when executed, cause the processor 502 to perform a safety warning method for the elderly based on non-invasive electricity behavior analysis.

[0151] The processor 502 provides computing and control capabilities to support the operation of the entire computer device 500.

[0152] The internal memory 504 provides an environment for the operation of the computer program 5032 in the non-volatile storage medium 503. When the computer program 5032 is executed by the processor 502, the processor 502 can execute a safety early warning method for the elderly based on non-invasive electricity behavior analysis.

[0153] This network interface 505 is used for network communication with other devices. Those skilled in the art will understand that... Figure 3 The structure shown is merely a block diagram of a portion of the structure related to the present application and does not constitute a limitation on the computer device 500 to which the present application is applied. The specific computer device 500 may include more or fewer components than those shown in the figure, or combine certain components, or have different component arrangements.

[0154] The processor 502 is used to run a computer program 5032 stored in the memory to implement all the steps of the elderly safety early warning method based on non-invasive electricity behavior analysis.

[0155] It should be understood that in the embodiments of this application, the processor 502 may be a central processing unit (CPU), or it may be other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor may be a microprocessor or any conventional processor.

[0156] It will be understood by those skilled in the art that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The computer program includes program instructions and can be stored in a storage medium, which is a computer-readable storage medium. The program instructions are executed by at least one processor in the computer system to implement the process steps of the embodiments of the above methods.

[0157] Therefore, the present invention also provides a storage medium. This storage medium can be a computer-readable storage medium. The storage medium stores a computer program, wherein when executed by a processor, the computer program causes the processor to perform all the steps of the elderly safety early warning method based on non-invasive electricity behavior analysis.

[0158] The storage medium can be any computer-readable storage medium capable of storing program code, such as a USB flash drive, portable hard drive, read-only memory (ROM), magnetic disk, or optical disk.

[0159] Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of both. To clearly illustrate the interchangeability of hardware and software, the components and steps of the various examples have been generally described in terms of functionality in the foregoing description. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementations should not be considered beyond the scope of this invention.

[0160] In the embodiments provided by this invention, it should be understood that the disclosed systems and methods can be implemented in other ways. For example, the system embodiments described above are merely illustrative. For example, the division of each unit is only a logical functional division, and there may be other division methods in actual implementation. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed.

[0161] The steps in the method of this invention can be adjusted, merged, or reduced in order according to actual needs. The units in the system of this invention can be merged, divided, or reduced according to actual needs. Furthermore, the functional units in the various embodiments of this invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit.

[0162] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a 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 all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, a terminal, or a network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present invention.

[0163] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any person skilled in the art can easily conceive of various equivalent modifications or substitutions within the technical scope disclosed in the present invention, and these modifications or substitutions should all be covered within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. A safety early warning method for the elderly based on non-invasive electricity behavior analysis, characterized in that, include: Obtain electricity consumption data from the smart meters in the residences of elderly people living alone to obtain initial data; The initial electricity consumption data were analyzed in depth to determine the elderly person's daily routine, activity level, and characteristics of commonly used electrical appliances, and to establish a personalized electricity consumption behavior model. Obtain real-time electricity consumption data from the smart meters in the residences of elderly people living alone; Real-time electricity consumption data is compared with the electricity consumption behavior model to identify high-risk abnormal patterns and obtain the degree of abnormality; wherein, the high-risk abnormal patterns include at least one of the following: no activity, disappearance of rhythm, long-term absence from home, and abnormal increase in nighttime activity; Different levels of early warning information are generated based on the degree of abnormality, triggering different intervention measures ranging from family reminders to emergency responses.

2. The elderly safety early warning method based on non-invasive electricity behavior analysis according to claim 1, characterized in that, The initial electricity consumption data is analyzed in depth to determine the elderly person's circadian rhythm, activity level, and characteristics of commonly used electrical appliances, and to establish a personalized electricity consumption behavior model, including: The initial electricity consumption data is analyzed in depth. The wake-up time is determined by analyzing the first continuous electricity consumption activity in the morning, the sleep time is inferred based on the low electricity consumption at night, and the peak use of electrical appliances during mealtimes is identified to understand the daily routine of the elderly. Calculate the average power or electricity consumption during a specific time period during the day to assess the intensity of activities, and determine whether daily life is normal by monitoring the stability of the base load; Identify the electrical appliances and their characteristics commonly used by the elderly, and construct personalized electricity usage behavior models.

3. The elderly safety early warning method based on non-invasive electricity behavior analysis according to claim 2, characterized in that, The electrical appliance includes at least one of an electric kettle, a television, and a rice cooker.

4. The elderly safety early warning method based on non-invasive electricity behavior analysis according to claim 1, characterized in that, The step of comparing real-time electricity consumption data with the electricity consumption behavior model to identify high-risk abnormal patterns and determine the degree of abnormality includes: The real-time electricity consumption data is input into the electricity consumption baseline profile model for analysis, in order to compare the real-time electricity consumption data with the work and rest patterns, activity levels and equipment usage patterns in the electricity consumption baseline profile model, identify deviations, and determine the degree of deviation. Detect abnormal patterns in wakefulness, circadian rhythm, outdoor activities, and nighttime activities to determine if there are any risks, and identify abnormal patterns; The degree of anomaly is determined based on the anomaly pattern and the degree of deviation.

5. The elderly safety early warning method based on non-invasive electricity behavior analysis according to claim 4, characterized in that, The abnormal modes include at least one of the following: no activity abnormal mode, rhythm disappearance abnormal mode, prolonged absence from home abnormal mode, and abnormal increase in nighttime activity abnormal mode. The no activity abnormal mode refers to the absence of the expected peak electricity consumption after waking up; the rhythm disappearance abnormal mode refers to the lack of normal daily rhythm characteristics in the electricity consumption curve throughout the day; the prolonged absence from home abnormal mode refers to the electricity load remaining at the baseline level for a long period of time, showing no signs of human activity; and the abnormal increase in nighttime activity abnormal mode refers to the frequency or duration of nighttime activities being higher than usual.

6. The elderly safety early warning method based on non-invasive electricity behavior analysis according to claim 1, characterized in that, The generation of different levels of early warning information based on the degree of abnormality triggers different intervention measures, ranging from family reminders to emergency responses, including: The warning level is automatically determined based on the degree of abnormality. Based on the aforementioned warning level, targeted warning information is generated, triggering different intervention measures ranging from family reminders to emergency responses.

7. The elderly safety early warning method based on non-invasive electricity behavior analysis according to claim 6, characterized in that, Based on the aforementioned warning level, targeted warning information is generated, triggering different intervention measures ranging from family reminders to emergency responses, including: Based on the warning level, targeted warning information is generated. When the warning level is low, a gentle reminder is sent to community volunteers or children through the application, suggesting that they pay attention to the elderly's living conditions. When the warning level is medium to high, a work order is created and the community grid worker or elderly care consultant is notified to contact the elderly by phone or visit them at home, and feedback on the handling is requested. When the warning level is an emergency alarm level, other sensors are linked to notify emergency contacts and the community emergency service department.

8. A safety early warning method for the elderly based on non-invasive electricity behavior analysis, characterized in that, include: The first acquisition unit is used to acquire relevant electricity consumption data of the smart meter in the residence of the elderly living alone in order to obtain initial data; A unit is established to perform in-depth analysis of the initial electricity consumption data, determine the elderly person's daily routine, activity level, and characteristics of commonly used electrical appliances, and establish a personalized electricity consumption behavior model. The second acquisition unit is used to acquire real-time electricity consumption data of the smart meter in the residence of the elderly living alone. An anomaly identification unit is used to compare real-time electricity consumption data with the electricity consumption behavior model to identify high-risk anomaly patterns in order to obtain the degree of anomaly; wherein, the high-risk anomaly pattern includes at least one of no activity, disappearance of rhythm, prolonged absence from home, and abnormal increase in nighttime activity; The early warning unit is used to generate different levels of early warning information based on the degree of abnormality, triggering different intervention measures ranging from family reminders to emergency responses.

9. The elderly safety early warning system based on non-invasive electricity behavior analysis according to claim 8, characterized in that, The establishment unit includes: The analysis subunit is used to perform in-depth analysis of the initial electricity consumption data, determine the wake-up time by analyzing the first continuous electricity consumption activity in the morning, infer the fall-up time based on the low electricity consumption at night, and identify the peak use of electrical appliances during mealtimes to understand the daily routine of the elderly. The calculation subunit is used to calculate the average power or electricity consumption during a specific time period of the day to assess the intensity of activities, and to determine whether daily life is normal by monitoring the stability of the base load. The feature recognition subunit is used to identify the electrical appliances and their characteristics that elderly people frequently use, and to build a personalized fingerprint of their electricity usage behavior.

10. The elderly safety early warning method based on non-invasive electricity behavior analysis according to claim 8, characterized in that, The anomaly detection unit includes: The deviation identification subunit is used to input the real-time electricity consumption data into the electricity consumption baseline profile model for analysis, so as to compare the real-time electricity consumption data with the work and rest patterns, activity levels and equipment usage patterns in the electricity consumption baseline profile model, identify deviations, and determine the degree of deviation. The pattern detection subunit is used to detect abnormal patterns of wake-up, circadian rhythm, going out, and nighttime activities to determine whether there is a risk and to identify abnormal patterns. The degree determination subunit is used to determine the degree of anomaly based on the anomaly pattern and the degree of deviation.

Images