Intelligent management system for community home-based care service for aged

Abstract

The invention relates to the field of intelligent management systems, and discloses an intelligent management system for community home-based care service, which comprises a data acquisition module, a space-time correction module, a data fusion module, an anomaly detection and feedback module, an intelligent service scheduling module and a user application layer, the data acquisition module is used for acquiring health, behavior and environment data of the old people in real time through various sensors; the space-time correction module ensures the space-time consistency of the data; the data fusion module adopts a dynamic weighted fusion algorithm to improve the precision and robustness of multi-source data; the abnormity detection and feedback module identifies abnormity and adjusts a strategy based on a machine learning algorithm; the intelligent service scheduling module dynamically adjusts service requirements and resource allocation according to the real-time data; and the user application layer provides real-time monitoring, alarming and service management. Compared with the prior art, the method has remarkable advantages in the aspects of data processing precision, anomaly detection flexibility and service scheduling efficiency.

Description

Technical Field

[0001] The present invention relates to the technical field of intelligent management systems, and in particular to an intelligent management system for community home-based elderly care services. Background Art

[0002] With the increasing aging of society, community home-based elderly care services have gradually become one of the important forms of elderly care services. By installing sensor technologies such as health monitoring equipment and environmental sensing devices, community home-based elderly care systems can collect the health status, behavior data and living environment parameters of the elderly in real time, and provide health management, abnormal alarms and service scheduling functions for the elderly. However, the existing community home-based elderly care service systems still have the following deficiencies in technical implementation.

[0003] First, in terms of data collection and processing, the data interaction of multiple sensors in the existing technology has obvious time synchronization errors and spatial deviation problems. Due to factors such as the hardware characteristics, deployment locations, and communication delays of different sensors, the data collected by the sensors lack temporal and spatial consistency. Most current systems only rely on simple rule corrections or fixed timestamp adjustment methods, which makes it difficult to achieve accurate alignment of multi-source data in complex scenarios. This data processing method performs poorly when faced with dynamic changes in sensors, resulting in large errors in the subsequent data fusion stage.

[0004] Secondly, in terms of multi-source data fusion, existing technologies usually use fixed weight allocation or weighted average algorithm to simply merge sensor data. This method cannot dynamically adjust weights to adapt to changes in sensor data quality, nor can it effectively handle noise and anomalies in multi-source data. Especially in complex sensor networks, due to the heterogeneity and uncertainty of data, the results of traditional fusion methods are low in accuracy and lack of robustness, which cannot meet the needs of elderly care service systems for high-precision perception results.

[0005] Thirdly, in terms of anomaly detection and feedback mechanisms, most existing technologies use fixed thresholds or single rule algorithms for anomaly identification. This approach lacks flexibility and intelligence, and is difficult to handle diverse anomaly patterns and complex scenarios. For example, when there is a potential correlation between environmental parameters and health data, fixed threshold algorithms often fail to accurately identify anomalies. In addition, existing anomaly feedback mechanisms usually only provide simple alarms or sensor shutdown operations, lack the ability to dynamically adjust system parameters, and cannot achieve efficient responses to abnormal events.

[0006] Finally, in terms of service scheduling and resource allocation, existing technologies mainly rely on static rules or manual scheduling methods, which makes it difficult to dynamically generate service needs and match optimal resources based on the real-time changing health status and environmental information of the elderly. This approach leads to inefficient service response, especially in multi-task and multi-resource scenarios, where traditional scheduling systems have difficulty coordinating task priorities and path planning, and cannot fully utilize service resources, limiting the improvement of service quality.

[0007] Therefore, the present invention proposes an intelligent management system for community home-based elderly care services to address the deficiencies of the prior art. Summary of the invention

[0008] In view of the shortcomings of the prior art, the present invention provides an intelligent management system for community home-based elderly care services. The system solves the shortcomings of traditional elderly care service systems in data collection, processing, anomaly detection, resource scheduling, etc. by integrating intelligent processing and analysis of multi-source sensor data.

[0009] To achieve the above objectives, the present invention is implemented through the following technical solutions: an intelligent management system for community home-based elderly care services, comprising:

[0010] A data acquisition module, used to collect data from multiple source sensors;

[0011] The time-space correction module is used to perform time synchronization and spatial position correction on the collected multi-source data to eliminate the time deviation and space deviation between sensors;

[0012] The data fusion module is used to dynamically fuse the multi-source data after spatiotemporal correction to generate global perception results;

[0013] The anomaly detection and feedback module is used to detect anomalies in the data fusion results and dynamically adjust the sampling strategy and weight distribution of the sensor according to the detection results;

[0014] Intelligent service scheduling module, used to generate service requirements based on global perception results, and perform resource matching and scheduling;

[0015] The user application layer is used to provide real-time monitoring data, abnormal alarm information and service management functions to users and service executors.

[0016] Preferably, the spatiotemporal correction module comprises:

[0017] The time synchronization unit is used to dynamically update the sensor timestamp through the time interaction of adjacent nodes;

[0018] The spatial alignment unit is used to dynamically adjust the spatial coordinates of the sensors according to the relative positions and expected distances between the sensors.

[0019] Preferably, the data fusion module performs data fusion based on a Kalman filter method with dynamic weight allocation, wherein:

[0020] The gain coefficient is calculated by predicting the deviation between the prior value of the target state and the current observed value;

[0021] The target state is updated according to the gain coefficient to reduce the impact of measurement noise on the fusion result.

[0022] Preferably, the dynamically adjusting the sampling strategy and weight distribution of the sensor specifically includes:

[0023] Calculate the error between the sensor observation and the fusion estimate;

[0024] The weights are dynamically adjusted according to the errors so that sensors with high credibility contribute more to the fusion results.

[0025] Preferably, the anomaly detection and feedback module detects abnormal observations by calculating the statistical distance of multi-source data, and includes the following steps:

[0026] Construct the mean and covariance matrix of the fused data;

[0027] Calculate the degree of deviation between the current observation value and the center of the data distribution;

[0028] If the degree of deviation exceeds the preset threshold, it is marked as an outlier and the sampling frequency or weight of the corresponding sensor is adjusted.

[0029] Preferably, the step of adjusting the sampling frequency or weight of the corresponding sensor includes:

[0030] When the data collected by the sensor is frequently judged to be abnormal, the sampling frequency of the sensor is reduced;

[0031] When the abnormality of sensor data is alleviated, the sampling frequency is gradually restored or the weight is increased.

[0032] Preferably, the intelligent service scheduling module includes:

[0033] Demand identification unit, used to generate service demands of the elderly based on global perception results, including health monitoring, medical assistance and environmental adjustment;

[0034] A resource matching unit, used to select the optimal service resource according to service requirements;

[0035] The task allocation unit is used to allocate tasks and determine the optimal service path and time through optimization algorithms.

[0036] Preferably, the user application layer provides the user with:

[0037] Real-time monitoring information of health parameters of the elderly;

[0038] Visual data of the environment status;

[0039] Alarm prompts and history query functions for abnormal events.

[0040] Preferably, the data collected by the data collection module includes:

[0041] Physiological data: including but not limited to heart rate, blood pressure, and body temperature;

[0042] Behavioral data: including but not limited to movement trajectory and fall events;

[0043] Environmental data: including but not limited to temperature and humidity, air quality, and smoke concentration.

[0044] Preferably, the system optimizes the collaborative work of multiple modules through a feedback closed loop, including:

[0045] Dynamic parameter adjustment of data acquisition module;

[0046] Adaptive convergence of spatiotemporal correction modules;

[0047] Dynamic weight optimization of data fusion module;

[0048] Real-time update of intelligent service scheduling module.

[0049] The present invention provides an intelligent management system for community home-based elderly care services. It has the following beneficial effects:

[0050] 1. The present invention adopts a technical solution combining multi-sensor data interaction with a spatiotemporal correction module. Through time synchronization correction and spatial position alignment, it ensures that the data collected by different sensors can be fused and processed in a unified spatiotemporal reference system, achieving the technical effect of improving the processing accuracy and perception consistency of multi-source data. Compared with the technical solution of using a single sensor or independent processing of sensor data in the prior art, it solves the data distortion problem caused by time deviation and spatial inconsistency between sensors, and effectively improves the system's adaptability in a dynamic environment.

[0051] 2. The present invention adopts a multi-layer data fusion algorithm based on dynamic weight allocation, combined with a weighted average method and Bayesian filtering technology, to efficiently fuse and suppress noise from multi-source data, achieving the technical effect of generating high-precision global perception results. Specifically, the weight allocation can be dynamically adjusted according to the quality, noise characteristics and historical reliability of the sensor data, further enhancing the credibility of the fusion results. Compared with the data fusion method using fixed weights or simple weighted average in the prior art, it solves the problem of insufficient adaptation caused by the heterogeneity of sensor data and improves the robustness and real-time performance of the fusion algorithm.

[0052] 3. The present invention adopts an anomaly detection and feedback mechanism that combines a rule engine, a statistical method, and a machine learning model, which can accurately identify anomalies in multi-source sensor data in real time, and dynamically adjust the sampling strategy and weight distribution according to the type of anomaly, thereby achieving the technical effect of enhancing the robustness of the system and the ability to cope with anomalies. For example, the rule engine can trigger adaptive responses based on different types of anomalies, and the machine learning model can handle complex nonlinear anomaly patterns through training and iterative optimization. Compared with the detection scheme in the prior art that only relies on fixed thresholds or simple anomaly classification algorithms, it solves the problems of insufficient accuracy in recognizing complex anomaly patterns and the inability of the system to dynamically adjust according to the environment.

[0053] 4. The present invention adopts an intelligent service scheduling module based on multi-objective optimization. Through the coordinated optimization of demand identification, resource matching and task allocation, it realizes the efficient utilization of service resources and real-time dynamic scheduling in multi-task scenarios, and achieves the technical effect of improving service response speed and accuracy. For example, a genetic algorithm is used to optimize resource allocation, and the shortest path task scheduling is realized through a path planning algorithm, which further reduces the time cost of service response. Compared with the static rule-driven scheduling method in the prior art, it solves the problems of low resource allocation efficiency, poor flexibility in service path planning, and insufficient adaptability to complex task scenarios, and significantly improves the overall performance of service scheduling. BRIEF DESCRIPTION OF THE DRAWINGS

[0054] Figure 1 This is an architecture diagram of an intelligent management system for community home-based elderly care services. DETAILED DESCRIPTION

[0055] The technical solution of the present invention will be described clearly and completely below in conjunction with the accompanying drawings of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0056] Please see attached Figure 1 The embodiment of the present invention provides an intelligent management system for community home-based elderly care services. The following is a detailed description of each module of the present invention.

[0057] Data acquisition module, used to collect data from multiple source sensors

[0058] The data acquisition module of the present invention is used to obtain relevant data of the elderly and their living environment from multi-source sensors as the basis for subsequent spatiotemporal correction, data fusion and service scheduling. The data acquisition module relies on various types of sensors and realizes data transmission and preliminary processing through standard communication protocols to ensure the integrity and real-time nature of the data.

[0059] Generally speaking, the data acquisition module is mainly responsible for collecting the following types of data:

[0060] Physiological data of the elderly, including health indicators such as heart rate, blood pressure, and body temperature;

[0061] Behavioral data of the elderly, including movement trajectory, activity intensity, fall events, etc.;

[0062] Living environment data, including indoor temperature and humidity, air quality (such as PM2.5 concentration), smoke concentration, etc.

[0063] As an option, the data collection module can perform preliminary screening of the collected data through preprocessing methods to eliminate missing values, extreme values ​​or duplicate values ​​to ensure the basic reliability of the data.

[0064] In some embodiments of the present invention, the data acquisition module includes the following main components:

[0065] Sensor network: It is composed of distributed sensor nodes covering the target area. Sensors include wearable devices (such as smart bracelets), environmental sensors (such as temperature and humidity sensors, air quality detectors) and behavioral sensors (such as cameras, accelerometers).

[0066] Data collection unit: responsible for collecting the data generated by the sensor nodes and transmitting them to the central server at predetermined time intervals.

[0067] Communication module: Use standard communication protocols (such as MQTT, CoAP, HTTP) to achieve two-way data transmission between sensors and servers.

[0068] Specifically, the implementation of the data acquisition module includes the following technical features:

[0069] In one possible implementation, the collection of physiological data is completed by a smart bracelet worn on the wrist of the elderly. The built-in optical heart rate sensor and blood pressure detection module of the smart bracelet can measure the heart rate and blood pressure data of the elderly at fixed intervals (such as 1 minute) and generate a data packet with a timestamp.

[0070] As an option, the smart bracelet can simultaneously record the number of steps, exercise intensity, and sleep duration, and transmit the data to the gateway device via Bluetooth or Wi-Fi. The gateway device further uploads the data to the central server.

[0071] Generally, behavioral data is collected through cameras and accelerometers installed in the room. The camera can capture the activity trajectory of the elderly, while the accelerometer can detect falls. The working principle of the accelerometer is as follows:

[0072] The sensor records the three-axis acceleration data in real time. x ,a y ,a z ;

[0073] By calculating the modulus of acceleration Determine whether there are sudden changes;

[0074] If |a| exceeds a preset threshold (such as 2g, where g is the acceleration of gravity), it is determined as a fall event and an alarm message is generated.

[0075] In environmental data collection, in some embodiments, the temperature and humidity sensor is used to record indoor temperature and humidity data. Specifically, the sensor regularly measures the current environmental temperature and humidity and generates formatted time series data. For example, the data format collected by a temperature and humidity sensor is as follows:

[0076] Data packet = {timestamp: t, temperature: T, humidity: H}

[0077] Where t is the acquisition time, T and H are the current temperature and humidity respectively.

[0078] As an extension, the air quality detector can simultaneously record changes in PM2.5 concentration, carbon dioxide concentration, and harmful gas (such as formaldehyde) concentration. The detection data is uploaded in real time through the built-in Wi-Fi module to form a complete environmental monitoring sequence.

[0079] In another possible implementation, in order to improve the reliability of data transmission, the communication module uses the MQTT protocol for data transmission. The advantages of the MQTT protocol are lightweight design, support for efficient data transmission in low-bandwidth network environments, and high anti-interference ability. The communication module can also use the QoS (Quality of Service) mechanism to dynamically adjust the transmission priority according to the importance of the data. For example, health data and fall alarm information have higher priority than environmental monitoring data.

[0080] In the data acquisition module, the data preprocessing unit performs basic screening on the collected data to ensure data quality. Specifically, the unit can perform the following steps:

[0081] Check the completeness of the data and remove missing values;

[0082] By defining a physical reasonable range, we can eliminate extreme values. For example, data with a heart rate lower than 30 beats / minute or higher than 200 beats / minute is considered invalid data.

[0083] Detect duplicate packets and remove redundant records.

[0084] In this embodiment, the data acquisition module not only supports heterogeneous data acquisition of multiple types of sensors, but also ensures the reliability and real-time performance of the data through flexible communication mechanisms and preprocessing algorithms. The collected data will be passed to the spatiotemporal correction module as the basis for subsequent processing.

[0085] The time-space correction module is used to synchronize the time and correct the spatial position of the collected multi-source data to eliminate the time and space deviations between sensors.

[0086] The spatiotemporal correction module in the present invention is mainly used to solve the problem of possible deviations in time and space of multi-source sensor data. Due to factors such as different working clocks, deployment location differences, and signal propagation delays between sensors, there are errors in their time synchronization and spatial alignment. The task of the spatiotemporal correction module is to eliminate these errors through technical means such as time synchronization and spatial correction, so that subsequent data fusion processing can obtain accurate global perception results.

[0087] In this embodiment, the spatiotemporal correction module includes two submodules, time synchronization and space alignment, which are responsible for eliminating time deviation and space deviation respectively. By coordinating the timestamps and location coordinates of each sensor, this module provides the data fusion module with corrected high-quality input to ensure that subsequent analysis, judgment and service scheduling are based on accurate perception data.

[0088] Generally speaking, the workflow of the spatiotemporal correction module is to first correct the clocks of each sensor through time synchronization so that their timestamps are consistent; then, through the spatial alignment algorithm, the spatial coordinates of the sensor are corrected to eliminate the spatial errors caused by differences in the deployment positions of the sensors.

[0089] Specifically, in some embodiments, the time synchronization function of the time-space correction module adopts a synchronization method based on adjacent nodes. The specific method is that the sensor synchronizes based on the network topology relationship by exchanging timestamp information with its adjacent nodes. The process includes the following steps:

[0090] Each sensor node first records the local timestamp t i , and sends it to its neighboring nodes through the communication network. After receiving the time information, the neighboring node records the receiving time t of the data packet i , and calculate the synchronization difference between the two based on the time difference information;

[0091] Each sensor node uses a synchronization algorithm to adjust its local clock based on the feedback information from neighboring nodes. Common synchronization methods include the Network Time Protocol (NTP) and algorithms based on synchronization protocols (such as delay estimation methods). Assume that the delay between sensors i and j is δ ij , the clock synchronization algorithm of sensor i can be expressed as:

[0092]

[0093] Among them, α is the adjustment coefficient, which represents the synchronization accuracy.

[0094] As an option, the spatiotemporal correction module can adopt a distributed time synchronization method, introduce a timestamp correction strategy between sensor nodes, and iteratively optimize the relative time difference of each sensor data so that the timestamps of all sensors finally converge to a consistent global time.

[0095] In terms of spatial alignment, the correction method in this embodiment is based on the relative position relationship of the sensor nodes and uses a spatial geometry algorithm to dynamically adjust the position coordinates of the sensors. Assume that the relative distance between sensors i and j is d ij ,In the initial stage, the system will establish a preliminary spatial coordinate system based on ,the installation locations of sensors (such as sensors fixed in the room) and the known ,environmental data.

[0096] Specifically, in one possible implementation, spatial correction uses the following optimization method to minimize the sensor node position deviation:

[0097] By constraining the relative distance between sensors, the multi-sensor positioning equation is established. Assume that the coordinates of sensors i and j are (x i ,y i ) and (x j ,y j ). Its relative distance is d ij , then:

[0098]

[0099] The system adjusts the spatial coordinates of each sensor to minimize the calculated actual distance and the expected relative distance. The optimal coordinate adjustment is solved by the least squares method (or other optimization algorithms):

[0100]

[0101] in, is the known expected distance, and X is the sensor coordinate set to be optimized.

[0102] As an extension, spatial correction can also be combined with dynamic information from sensors (such as displacement data from accelerometers) for iterative optimization. By combining different data sources, the spatiotemporal correction module can adaptively adjust the correction strategy in different environments and scenarios to improve correction accuracy.

[0103] In addition, in order to improve the efficiency of spatiotemporal correction, the spatiotemporal correction module can also adopt a hierarchical synchronization and spatial correction strategy according to the network topology and the working status of the sensor.

[0104] Specifically, for sensors in the same physical area, local correction can be performed, while for sensors across regions, global correction is performed through a multi-hop collaborative synchronization method. This hierarchical correction method can effectively reduce the computational complexity while maintaining high accuracy.

[0105] In some embodiments, the time synchronization and space alignment processes of the spatiotemporal correction module can be performed in parallel to improve the correction efficiency of the system. For example, after the time synchronization is completed, the sensor clock is adjusted to a globally consistent standard time, and the space alignment algorithm is adjusted in real time based on the initial position of the sensor to ensure the spatiotemporal consistency of the sensor data.

[0106] In general, the spatiotemporal correction module enables data from different sensors to be fused in the same spatiotemporal reference system through precise time synchronization and spatial correction. This not only effectively improves the accuracy of data processing, but also provides a reliable perception data foundation for subsequent data fusion, anomaly detection, and intelligent service scheduling modules.

[0107] The data fusion module is used to dynamically fuse the multi-source data after spatiotemporal correction to generate global perception results.

[0108] The data fusion module of the present invention is designed to process the spatiotemporal corrected data collected from multi-source sensors, and combine the data characteristics of different types of sensors to generate accurate global perception results. Since the sensors involved in the community home-based elderly care service system are diverse and widely distributed, the data fusion module not only needs to process data from health monitoring equipment, behavior monitoring sensors and environmental perception devices, but also needs to improve the accuracy, reliability and integrity of the data through a reasonable fusion algorithm.

[0109] In general, the data fusion module is based on the redundancy and complementarity of multi-sensor information. By weighted fusion of various types of data, it eliminates the errors that may be caused by a single sensor and provides more accurate and comprehensive perception results. The data fusion process can be divided into multiple stages, including preliminary fusion, weighted processing, outlier detection, and fusion result optimization.

[0110] In some embodiments, the workflow of the data fusion module is as follows: first, the data from different sensors have been corrected in time and space through the space-time correction module; then, the results of the preliminary fusion will be passed to the weighted processing unit, and different weights will be assigned according to the data quality and reliability of each sensor; finally, the data after outlier detection and optimization will be provided to the subsequent modules of the system (such as the intelligent service scheduling module) as the global perception result.

[0111] Specifically, the data fusion module in this embodiment adopts a multi-level data fusion algorithm based on weighted average and Bayesian filtering. The weighted average algorithm and Bayesian filtering can effectively combine information from multiple data sources to ensure that the final perception result is more accurate and reliable.

[0112] In multi-source data fusion, the signals of different sensors are usually processed by weighting. For example, suppose there are data x from different sensors. 1 ,x 2 ,…,x n , and the corresponding weight w 1 ,w 2 ,…,w n , the weighted average can be calculated by the following formula:

[0113]

[0114] Among them, x fused is the final fusion result, weight w i The sensors are allocated based on their reliability, accuracy and historical performance. In this way, the data fusion module can combine the data from multiple sensors according to the weight of each sensor and eliminate the errors that may be caused by a single data source.

[0115] Bayesian filtering is another commonly used data fusion method, especially suitable for data that needs to handle uncertainty and dynamic changes. Bayesian filtering gradually updates the state estimate by combining prior information and new observation data. In this embodiment, assuming that the system needs to fuse data from health monitoring sensors for the elderly (such as heart rate sensors and blood pressure sensors) and environmental sensors (such as temperature and humidity sensors), the Bayesian filtering process can be expressed as:

[0116]

[0117] Among them, P(x k |y 1 ,y 2 ,…,y k ) is the posterior probability, indicating that given the observed data y 1 ,y 2 ,…,y kAfter that, the system state x k The probability of P(y k |x k ) is the likelihood function, indicating that in state x k Observe the data y k The probability of P(x k |y 1 ,y 2 ,…,y k-1 ) is the prior probability, indicating that given the previous observation data, the system state x k probability.

[0118] Through Bayesian filtering, the data fusion module can dynamically adjust and optimize the global perception results based on the current data provided by the sensor and the previous fusion results, especially with high robustness and adaptability in dynamically changing environments.

[0119] As an extension, the data fusion module can also combine the advantages of weighted averaging and Bayesian filtering to adopt a hybrid algorithm. In this hybrid algorithm, the high-precision, low-noise sensor data is first fused by weighted averaging, and then the low-precision, noisy data is corrected and optimized by applying Bayesian filtering. In this way, the system can maximize the accuracy and stability of data fusion while ensuring computational efficiency.

[0120] In addition, the data fusion module also includes anomaly detection and optimization submodules. In general, the anomaly detection module monitors the data of each sensor by setting thresholds and anomaly detection algorithms (such as statistical Z-score methods or machine learning-based models) to detect anomalies in the data. For example, in health monitoring data, a heart rate exceeding 200 beats / minute or less than 30 beats / minute may be abnormal data; in environmental monitoring data, a temperature exceeding the set safety range should also be considered abnormal. Abnormal data will be marked and weighted during the fusion process to avoid affecting the final perception results.

[0121] In one possible implementation, the anomaly detection and optimization module further optimizes the weight distribution of sensors through dynamic adjustment based on adaptive thresholds. Specifically, if the data of a sensor fluctuates greatly, resulting in poor stability of its fusion result, the weight of the sensor will be automatically reduced in the subsequent data fusion process.

[0122] In this embodiment, the data fusion module can not only ensure the efficient fusion of multi-sensor data through multi-level algorithms and optimization mechanisms, but also adaptively adjust data weights in a dynamically changing environment to provide accurate global perception results in real time. The accuracy and robustness of data fusion provide reliable input for subsequent anomaly detection, intelligent service scheduling, and user feedback modules, further enhancing the overall efficiency and stability of the system.

[0123] The anomaly detection and feedback module is used to detect anomalies in the data fusion results and dynamically adjust the sampling strategy and weight distribution of the sensor according to the detection results.

[0124] The anomaly detection and feedback module in the present invention is mainly responsible for real-time monitoring and anomaly detection of the fused multi-source sensor data to ensure the accuracy and reliability of the system's perception results. By continuously tracking and monitoring the data stream, this module can identify any abnormal behavior or state changes that do not meet expectations, and promptly feedback to the system for corresponding adjustments. The core goal of this module is to provide an intelligent, accurate, and real-time anomaly identification mechanism to avoid improper judgments or service responses caused by abnormal data.

[0125] In general, the anomaly detection and feedback module adopts a comprehensive solution based on statistical methods, rule engines, and machine learning algorithms, and combines multiple technical means to ensure efficient anomaly identification in various scenarios. The key task of anomaly detection is to identify and process outliers, fluctuation anomalies, pattern mutations, etc. in the data, and dynamically adjust the sampling strategy and weight of sensor data according to the detected anomaly type to ensure that the subsequent data processing process will not be disturbed by errors.

[0126] In some embodiments, the anomaly detection and feedback module first performs a preliminary detection on the sensor data using a statistical method, using common anomaly detection algorithms such as Z-score or IQR (interquartile range) method to quickly identify outliers in the data. For example, given a set of sensor data x 1 ,x 2 ,…,x n , we can calculate the mean μ and standard deviation σ of the data set, and then calculate the Z value of each data point according to the Z-score algorithm:

[0127]

[0128] If | Z i |>α (where α is the preset threshold), then x i It is judged as an outlier. Generally speaking, when the Z value is greater than 3, the data point can be considered an outlier.

[0129] Specifically, in this embodiment, the Z-score method is used to detect abnormal data generated by health monitoring sensors (such as heart rate, blood pressure, etc.). Under normal circumstances, the heart rate of the elderly should be kept within a healthy range (for example, 60 to 100 beats per minute). If the heart rate data at a certain moment exceeds this range, the Z-score method is used to detect that the data point is abnormal, and feedback is given based on the abnormal detection result to notify the subsequent processing module to perform special processing or delete the data point.

[0130] As an option, the anomaly detection module cannot use the interquartile range method (IQR) for outlier detection. By calculating the first quartile Q1 and the third quartile Q3 of the data set, and calculating the interquartile range IQR = Q3-Q1, the range of abnormal data is determined according to the formula:

[0131] Abnormal range = [Q1-1.5×IQR, Q3+1.5×IQR]

[0132] If the data point exceeds this range, it is considered abnormal data. In some specific cases, the IQR method can better handle non-normally distributed data, and is particularly suitable for data analysis of environmental perception sensors (such as temperature and humidity sensors).

[0133] If the data point exceeds this range, it is considered abnormal data. In some specific cases, the IQR method can better handle non-normally distributed data, and is particularly suitable for data analysis of environmental perception sensors (such as temperature and humidity sensors).

[0134] In one possible implementation, the anomaly detection in this embodiment is also combined with a detection method based on a rule engine to classify and process abnormal data. Assuming that the system detects an abnormal data point (such as a heart rate that is too high), the rule engine can determine whether the anomaly is a "health alert" or a "sensor failure" based on the rules. For example, if the heart rate data exceeds the set threshold and the sensor sampling frequency is too low, it may be a sensor failure; and if the heart rate data remains too high for a long time, it may be that the user has a health problem. The rule engine classifies the abnormal data according to preset rules and feeds the results back to subsequent modules.

[0135] Specifically, the workflow of the rule engine is as follows:

[0136] After receiving the data from the sensor, the potential abnormal data points are first screened out through an anomaly detection algorithm;

[0137] Match the abnormal data with rules and determine the abnormal type based on the preset rules. For example, a high heart rate may be an "acute health problem";

[0138] Depending on the type of anomaly, the system will adopt different response strategies, such as eliminating data, recalibrating sensors, or sending alarm notifications.

[0139] The machine learning algorithm is also an important component of the anomaly detection and feedback module of the present invention. In order to improve the intelligence level of the system, the anomaly detection module can use a support vector machine (SVM) or a deep learning model for anomaly detection. By training the model to classify normal and abnormal data, the machine learning method can continuously optimize the detection algorithm based on historical data and improve the ability to recognize complex abnormal patterns.

[0140] In this embodiment, the support vector machine (SWM) method is used for training and anomaly detection. Assume that the feature of the sensor data is x=(x 1 ,x 2 ,…,x n ), the SVM model obtains an optimal decision hyperplane f(x) through training:

[0141] f(x)=w T x+b

[0142] If the output of f(x) is greater than a certain threshold, the data is considered normal; if it is less than the threshold, it is considered abnormal. Through the feedback of the training data set, the SVM model can adaptively adjust the model parameters w and b to achieve abnormal identification of more complex data patterns.

[0143] In some cases, the anomaly detection and feedback module can also be updated in real time using online learning algorithms, especially when the system is deployed in an environment with dynamic changes. For example, when the ambient temperature and humidity change greatly, the system can dynamically update the anomaly detection rules based on the online learning algorithm to adapt to the changes in the environment in a timely manner.

[0144] The feedback mechanism is an important part of the anomaly detection and feedback module. When an abnormal situation is detected, the system not only needs to record and report the abnormality, but also needs to dynamically adjust the sensor's sampling strategy, weight distribution, etc. according to the type of abnormality. If a sensor failure is detected or there is a lot of noise in the data, the system can adjust the sampling frequency of the sensor or reduce its weight in the data fusion stage. Through this feedback mechanism, the system can adaptively adjust the working status of the sensor and improve the accuracy of the overall data processing.

[0145] In one possible implementation, the feedback mechanism of the anomaly detection module includes the following aspects:

[0146] Real-time alarm: Once abnormal data is detected, an alarm is immediately sent to the system management interface or user to notify service personnel to handle it;

[0147] Data correction: Correct or replace abnormal data by using the most recent historical data or data from other sensors;

[0148] Dynamic sensor adjustment: Adjust the sensor’s sampling strategy, accuracy and other parameters based on the type of anomaly detected.

[0149] In short, the anomaly detection and feedback module can detect and feedback anomalies in multi-source sensor data in real time and efficiently through the comprehensive use of statistical methods, rule engines and machine learning technologies, and dynamically adjust system operating parameters through feedback mechanisms to ensure that the system can provide accurate and stable global perception results under any circumstances.

[0150] Intelligent service scheduling module, used to generate service requirements based on global perception results, and perform resource matching and scheduling

[0151] The intelligent service scheduling module in the present invention is responsible for dynamically generating service requirements and achieving optimal resource allocation and task scheduling based on the multi-sensor data fusion results and the feedback from the anomaly detection module. The core task of the intelligent service scheduling module is to convert the perceived health status, environmental information and behavior patterns of the elderly into specific service requirements, and optimize the matching of service resources through intelligent algorithms to achieve efficient resource utilization and accurate service provision.

[0152] Generally speaking, the intelligent service scheduling module generates corresponding service types (such as medical assistance, environmental adjustment or psychological counseling) based on health monitoring, environmental perception and abnormal alarm information, and then matches the optimal service personnel or equipment through the optimization algorithm, and plans the service path and execution time to ensure the timeliness and efficiency of the service.

[0153] As an option, this module can combine technologies such as genetic algorithms, heuristic algorithms, and path planning algorithms to solve the multi-task and multi-resource allocation problem through dynamic scheduling and real-time optimization mechanisms. Specifically, this module includes a demand identification unit, a resource matching unit, and a task allocation unit. Each unit cooperates with each other to complete the full-process scheduling from demand generation to service execution.

[0154] Requirements Identification Unit

[0155] In this embodiment, the demand identification unit first identifies the service demand through the global perception results and the abnormal detection results. Generally speaking, abnormal conditions in health monitoring data (such as high heart rate, low blood pressure) will trigger the need for medical assistance; environmental data (such as substandard air quality, abnormal indoor temperature and humidity) will trigger the need for environmental adjustment; behavioral data (such as fall detection, long-term inactivity) will trigger the need for emergency alarm.

[0156] In one possible implementation, the demand identification unit automatically generates service demands through a rule engine and a classification model. For example, when the heart rate of an elderly person exceeds a set threshold (such as 120 beats per minute), the system identifies a "health alert" and generates a medical service demand; when the indoor air quality index (AQI) exceeds a set range, the system identifies an "abnormal environment" and generates an air purification demand.

[0157] The rule formula of the requirement identification process can be described as:

[0158] D i =f(S i ,T i ,L i )

[0159] Among them, D i represents the i-th service demand, S i is the health data status, T i is the behavior mode state, L i is the environmental parameter state, and f is the rule mapping function.

[0160] Resource Matching Unit

[0161] In this embodiment, the resource matching unit selects the best service resource (such as service personnel, equipment or organization) according to the generated service demand. Generally, the resource matching unit needs to comprehensively consider factors such as the service personnel's skills, geographical location, current working status and response time to ensure the optimality of the matching result.

[0162] As an option, the resource matching unit uses a multi-objective optimization algorithm based on weight scoring for matching. Specifically, assuming that the system has m service requirements and n available resources, the objective function of resource matching is defined as:

[0163]

[0164] Among them, x ij is a decision variable, indicating whether the i-th demand matches the j-th resource (value is 1 or 0); ij It is the weight that comprehensively reflects the compatibility between service personnel and service needs.

[0165] In a possible implementation, the system completes resource matching through the following steps:

[0166] Calculate the distance between the service personnel's geographical location and the target location and use it as part of the weight of the fitness;

[0167] Adjust the weight distribution according to the service personnel's skill matching, task priority and availability;

[0168] Use heuristic algorithms (such as ant colony optimization or genetic algorithm) to find the optimal matching solution.

[0169] Task allocation unit:

[0170] The task allocation unit is responsible for planning the path and time of service execution based on the resource matching results and dynamically adjusting task scheduling. Specifically, the task allocation unit needs to solve the following two key problems:

[0171] Path planning: Generate the optimal access path for service personnel or equipment;

[0172] Time scheduling: Determine the priority and execution order of tasks.

[0173] In some embodiments, the task assignment unit uses a path planning technology based on the Dijkstra algorithm. Assuming that the service personnel need to visit multiple target points (such as the address of the elderly), the system first constructs a weighted graph G (V, E) between the target points, where V represents the set of target points and E represents the path weight (such as distance or time) between the points. Through the path planning algorithm, the task assignment unit can solve the shortest path or the optimal path.

[0174] As an alternative, the time scheduling problem can be solved by a linear programming model. Assume there are n tasks, each with an execution time of t i , and define the task priority p i The system is scheduled by minimizing the objective function of the total execution time T:

[0175]

[0176] And the following constraints need to be met:

[0177] The start time of all tasks must not be earlier than the current time;

[0178] Each task must be completed no later than the set deadline.

[0179] Feedback and Dynamic Scheduling

[0180] In this embodiment, the intelligent service scheduling module supports a real-time feedback mechanism. Generally, when unexpected situations occur during service execution (such as service personnel delays or equipment unavailability), the system can dynamically adjust task allocation and path planning based on real-time feedback. For example, when it is detected that a service personnel cannot arrive at the service location on time, the system can re-match backup resources or adjust task priorities.

[0181] As a possible implementation method, the feedback mechanism combines real-time location tracking technology and task status monitoring technology to ensure that the system can respond quickly to emergencies. By incorporating real-time feedback into the task scheduling process, the intelligent service scheduling module can effectively improve the system's robustness and service efficiency.

[0182] In summary, the intelligent service scheduling module realizes the precise scheduling and efficient execution of community home-based elderly care services through the multi-level design of demand identification, resource matching, task allocation and real-time feedback. The system combines rule engines and intelligent algorithms to not only meet diverse service needs, but also has a high degree of dynamic adaptability, providing reliable and convenient service support for the elderly.

[0183] User application layer, used to provide real-time monitoring data, abnormal alarm information and service management functions to users and service executors

[0184] The user application layer of the present invention is the core module for the interaction between the system and the end user. It is used to present the perception results after multi-source data processing to the user in an intuitive and friendly form, and provide users with real-time monitoring, abnormal alarm, service request and historical data query functions. The design of the user application layer is based on the actual needs of community home-based elderly care services, mainly for the elderly, their families and community managers, to ensure that users can easily obtain relevant information and perform necessary operations.

[0185] In general, the user application layer displays data to users through mobile terminals (such as mobile phone apps), desktop terminals (such as PC web pages) and community interactive terminals (such as smart screens). This module takes the results of multi-source data fusion as the core, and improves the ease of use of the system and the accuracy of information transmission through data visualization, event notification and interactive design.

[0186] As an option, the user application layer can also achieve remote interaction through intelligent voice assistants or instant messaging tools (such as SMS or push messages) to ensure the real-time information and convenience of operation.

[0187] Real-time monitoring interface

[0188] In this embodiment, the real-time monitoring interface of the user application layer is used to display the health status, activity information and environmental data of the elderly. Specifically, the health status includes data such as heart rate, blood pressure, and body temperature, the activity information includes the number of exercise steps, static time, etc., and the environmental data includes indoor temperature and humidity, air quality, and smoke alarm status.

[0189] Specifically, the real-time monitoring interface uses visual charts to intuitively display data change trends. For example, heart rate data can be presented in the form of a line chart to show the dynamic changes in the past hour; temperature and humidity data can be presented in the form of a bar chart to show the relationship between the current data and the comfort range.

[0190] In a possible implementation, the real-time monitoring interface also supports multi-layer data switching. Users can view more detailed sub-item data by clicking on a specific icon, such as viewing health monitoring records for a day, or switching to a historical curve of environmental monitoring data.

[0191] Abnormal alarm function

[0192] In this embodiment, the abnormal alarm function of the user application layer is used to send an alarm notification to the user when an abnormality is detected in the health status or environment of the elderly. Generally, the alarm information is highlighted in red and comes with detailed instructions and recommended operations.

[0193] As an option, the abnormal alarm function can be displayed in different levels according to the urgency of the abnormal event. For example, the first level alarm may be used for emergency events such as elderly people falling or high heart rate, and the system will immediately send a phone call or text message to notify family members or community workers; the second level alarm may be used for non-emergency events such as poor indoor air quality, and only the prompt information will be displayed on the user application layer interface.

[0194] Specifically, the notification content of the abnormal alarm function includes: event type (such as "fall detection"), time of occurrence, location and recommended actions (such as "Please contact emergency services immediately"). For example, when the system detects that the heart rate of an elderly person continues to exceed 120 beats per minute, the alarm message may be displayed as:

[0195] Title: Health abnormality alarm

[0196] Content: A high heart rate (125 beats / min) has been detected. Please take immediate action.

[0197] In one possible implementation, the user can customize the receiving mode and processing priority of the alarm notification. For example, the user can choose to only receive SMS notifications, or to send all alarm information to the mobile phones of family members first.

[0198] Service request and operation interface

[0199] In this embodiment, the user application layer provides a service request and operation interface, allowing users to submit specific service requests according to actual needs. Generally, users can select the required service type (such as medical assistance, psychological counseling, housekeeping services) through the interface and fill in relevant details (such as service time, location and remarks).

[0200] As an option, the service request interface can interact with the intelligent service scheduling module in real time to dynamically display available service resources. For example, after the user selects "housekeeping service", the interface will display a list of currently available service personnel and their estimated arrival time. After the user confirms the service request, the system will generate a task and assign it to the appropriate service resource.

[0201] Specifically, the operation process of the service request interface includes:

[0202] The user selects the type of service;

[0203] The system automatically recommends relevant service options based on current data;

[0204] Users fill in additional information for service requirements;

[0205] The system prompts the estimated service time and completes the request submission.

[0206] Historical data query and analysis

[0207] In this embodiment, the user application layer also supports historical data query and analysis functions, which facilitates users to view past health monitoring records and service history. For example, users can view the trend of heart rate data changes in the past week and analyze the changing patterns of their own health status.

[0208] As an option, the historical data query function supports data export and sharing. For example, users can generate a PDF report of their health data and send it to a medical institution or family member. The data report may include the following:

[0209] Overview of health status (such as average heart rate, blood pressure trend graph);

[0210] Detailed records of abnormal events (such as time of fall, number of environmental alarms);

[0211] System-generated health recommendations.

[0212] Specifically, the historical data analysis interface supports interactive operations, and users can quickly find the required data by dragging the timeline or adjusting the filter conditions. For example, users can choose to view only "abnormal events", and the system will display relevant records in the past period of time based on the filter conditions.

[0213] User personalization settings

[0214] In this embodiment, the user application layer supports personalized settings, allowing users to adjust the interface layout, notification method, and service preferences according to their own needs. In general, users can modify the following settings:

[0215] Notification method: select SMS, phone call, APP push and other notification methods;

[0216] Data display: adjust the priority and detail of data display;

[0217] Service Preferences: Set priorities for specific service personnel or service time periods.

[0218] As an extension, user personalization also supports multi-user mode. For example, community administrators can set the default preferences for public services through the system background, while family members can set personal notification and service preferences separately.

[0219] In summary, the user application layer provides users with a comprehensive interactive experience through real-time monitoring, abnormal alarm, service request and historical data query functions. This module combines an intuitive visual interface with flexible operation settings to ensure that users can easily obtain the required information and take quick action, providing reliable technical support for community home-based elderly care services.

[0220] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. An intelligent management system for community home-based elderly care services, characterized in that: include: A data acquisition module, used to collect data from multiple source sensors; The time-space correction module is used to perform time synchronization and spatial position correction on the collected multi-source data to eliminate the time deviation and space deviation between sensors; The data fusion module is used to dynamically fuse the multi-source data after spatiotemporal correction to generate global perception results; The anomaly detection and feedback module is used to detect anomalies in the data fusion results and dynamically adjust the sampling strategy and weight distribution of the sensor according to the detection results; Intelligent service scheduling module, used to generate service requirements based on global perception results, and perform resource matching and scheduling; The user application layer is used to provide real-time monitoring data, abnormal alarm information and service management functions to users and service executors.

2. The intelligent management system for community home-based elderly care services according to claim 1 is characterized in that: The spatiotemporal correction module comprises: The time synchronization unit is used to dynamically update the sensor timestamp through the time interaction of adjacent nodes; The spatial alignment unit is used to dynamically adjust the spatial coordinates of the sensors according to the relative positions and expected distances between the sensors.

3. The intelligent management system for community home-based elderly care services according to claim 1 is characterized in that: The data fusion module performs data fusion based on the Kalman filtering method of dynamic weight allocation, wherein: The gain coefficient is calculated by predicting the deviation between the prior value of the target state and the current observed value; The target state is updated according to the gain coefficient to reduce the impact of measurement noise on the fusion result.

4. The intelligent management system for community home-based elderly care services according to claim 3 is characterized in that: The dynamic adjustment of the sampling strategy and weight distribution of the sensor specifically includes: Calculate the error between the sensor observation and the fusion estimate; The weights are dynamically adjusted according to the errors so that sensors with high credibility contribute more to the fusion results.

5. The intelligent management system for community home-based elderly care services according to claim 1 is characterized in that: The anomaly detection and feedback module detects abnormal observations by calculating the statistical distance of multi-source data, and includes the following steps: Construct the mean and covariance matrix of the fused data; Calculate the degree of deviation between the current observation value and the center of the data distribution; If the degree of deviation exceeds the preset threshold, it is marked as an outlier and the sampling frequency or weight of the corresponding sensor is adjusted.

6. The intelligent management system for community home-based elderly care services according to claim 5, characterized in that: The step of adjusting the sampling frequency or weight of the corresponding sensor includes: When the data collected by the sensor is frequently judged to be abnormal, the sampling frequency of the sensor is reduced; When the abnormality of sensor data is alleviated, the sampling frequency is gradually restored or the weight is increased.

7. The intelligent management system for community home-based elderly care services according to claim 1 is characterized in that: The intelligent service scheduling module includes: Demand identification unit, used to generate service demands of the elderly based on global perception results, including health monitoring, medical assistance and environmental adjustment; A resource matching unit, used to select the optimal service resource according to service requirements; The task allocation unit is used to allocate tasks and determine the optimal service path and time through optimization algorithms.

8. The intelligent management system for community home-based elderly care services according to claim 1 is characterized in that: The user application layer provides users with: Real-time monitoring information of health parameters of the elderly; Visual data of the environment status; Alarm prompts and history query functions for abnormal events.

9. The intelligent management system for community home-based elderly care services according to claim 1, characterized in that: The data collected by the data collection module includes: Physiological data: including but not limited to heart rate, blood pressure, and body temperature; Behavioral data: including but not limited to movement trajectory and fall events; Environmental data: including but not limited to temperature and humidity, air quality, and smoke concentration.

10. The intelligent management system for community home-based elderly care services according to claim 1, characterized in that: The system optimizes the collaborative work of multiple modules through a feedback closed loop, including: Dynamic parameter adjustment of data acquisition module; Adaptive convergence of spatiotemporal correction modules; Dynamic weight optimization of data fusion module; Real-time update of intelligent service scheduling module.

Images