Time banking system for intelligent pension service, volunteer scheduling method, storage medium and electronic equipment

Abstract

The invention provides a time bank system for intelligent old-age care service, a volunteer scheduling method, a storage medium and electronic equipment, and relates to the technical field of Internet. The method comprises the steps of presetting coordinate parameters of an electronic geo-fence of a target service area, and generating corresponding access event data when a mobile terminal enters or leaves the electronic geo-fence; determining service start time and service end time of each wisher based on the in-out event data, and generating corresponding service duration data according to the service start time and the service end time; the feature extraction module is used for acquiring static information and dynamic service data related to candidate volunteers from a database, performing feature extraction processing on the information in combination with the service duration data, and generating feature vectors for matching; and calculating a matching score between the candidate volunteer and the target service object based on the feature vector and a preset machine learning model, and generating matching relation data based on the matching score.

Description

Technical Field

[0001] The present disclosure relates to the field of Internet technology, and in particular to a time bank system for intelligent elderly care services, a time bank volunteer scheduling method for intelligent elderly care services, a computer-readable storage medium, and an electronic device. Background Art

[0002] To alleviate the shortage of elderly care service personnel and improve service responsiveness and coverage, a growing number of cities are exploring the "time bank" model as a complementary approach to mutual-aid elderly care. This model promotes a reward system whereby time spent serving others can be used as a reserve for future services, offering scalability and social incentives. However, existing "time bank" systems generally lack intelligence and automation capabilities, failing to meet the demands of large-scale operations and service quality control.

[0003] The inventors discovered during their research that in existing time bank systems, the service hours of volunteers usually rely on manual sign-ins, offline records, or manual confirmation based on mobile applications. This type of method not only has human errors, but is also difficult to effectively prevent problems such as false reporting of services, falsification of hours, or omission of information in practice. Especially in unregulated places, relying solely on users clicking on "Start Service" and "End Service" operations can easily lead to data distortion and make it impossible to form a reliable basis for points. At the same time, in the process of recording service hours, the existing system lacks effective structured use of user location data, and is often unable to accurately determine whether the service actually occurs within the target service area, thus affecting the establishment of a closed service loop.

[0004] It should be noted that the information disclosed in the above background technology section is only used to enhance the understanding of the background of the present disclosure, and therefore may include information that does not constitute prior art known to ordinary technicians in the field. Summary of the Invention

[0005] The purpose of the embodiments of the present disclosure is to provide a time bank volunteer scheduling method, device, storage medium and electronic device for smart elderly care services, thereby at least to a certain extent solving the problem in related technologies that it is impossible to accurately determine whether the service actually occurs within the target service area, thereby affecting the establishment of a service closed loop.

[0006] According to a first aspect of an embodiment of the present disclosure, a time banking system for smart elderly care services is provided, comprising:

[0007] an electronic geofence detection module configured to pre-set coordinate parameters of an electronic geofence of a target service area and generate corresponding entry and exit event data when a mobile terminal enters or leaves the electronic geofence;

[0008] a service duration recording module, in communication connection with the electronic geofence detection module, configured to determine a service start time and a service end time of each volunteer based on the access event data, and generate corresponding service duration data accordingly;

[0009] a feature extraction module, connected with the service duration recording module, and establishing a data interaction channel with a preset volunteer information database, configured to acquire static information and dynamic service data related to the candidate from the database, and perform feature extraction processing on the information in combination with the service duration data, to generate a feature vector for matching;

[0010] a service matching module, connected with the feature extraction module, configured to calculate a matching score between the candidate volunteer and the target service object based on the feature vector and a preset machine learning model, and generate matching relationship data based on the matching score.

[0011] In an exemplary embodiment of the present disclosure, the electronic geofence detection module comprises:

[0012] a coordinate parameter receiving unit configured to load coordinate parameters of an electronic geofence corresponding to a target service area from a system preset configuration file, and store the coordinate parameters in a local buffer;

[0013] a position information acquisition unit, in communication connection with the coordinate parameter receiving unit, configured to acquire current position information in real time through a positioning component of a mobile terminal, and generate position information data for fence determination;

[0014] a fence determination unit, connected with the position information acquisition unit, configured to perform spatial comparison between the position information data and the coordinate parameters, to generate a determination identifier indicating an entering state or a leaving state;

[0015] an event generation unit, connected with the fence determination unit, configured to construct corresponding access event data based on the determination identifier, and transmit the access event data to the service duration recording module.

[0016] In an exemplary embodiment of the present disclosure, the position information acquisition unit generates position information data for fence determination in the following manner:

[0017] a global satellite positioning component of the mobile terminal is called to acquire current GPS position information data;

[0018] Wi-Fi access point information in the current environment is scanned, and corresponding Wi-Fi position information data is identified based on preset hotspot feature parameters;

[0019] Performing weighted fusion of the GPS location information data and the Wi-Fi location information data to generate fused location information data for fence determination;

[0020] A spatial position comparison operation is performed based on the fused position information data and the coordinate parameters of the electronic geographic fence to generate the determination identifier.

[0021] In an exemplary embodiment of the present disclosure, the service duration recording module includes:

[0022] a timestamp extraction unit, in communication with the electronic geofence detection module, for extracting, from the entry and exit event data, a first timestamp corresponding to the volunteer entering the electronic geofence and a second timestamp corresponding to the volunteer leaving the electronic geofence;

[0023] a time interval calculation unit, connected to the timestamp extraction unit, for calculating service time interval data based on a difference between the first timestamp and the second timestamp;

[0024] The duration generating unit is connected to the time interval calculating unit and is used to convert the service time interval data into service duration data in a standard format and transmit the service duration data to the feature extraction module.

[0025] In an exemplary embodiment of the present disclosure, the time banking system further includes a service quality analysis module; the service quality analysis module includes:

[0026] An image acquisition unit connected to the camera module of the mobile terminal and configured to periodically acquire video frame image data of the service scene during the volunteer service process;

[0027] a behavior recognition unit connected to the image acquisition unit, configured to analyze human motion features in the video frame image data based on a preset motion recognition model, and identify action labels including one or more service behaviors such as raising hands, feeding, supporting, turning over, using the toilet, cleaning, and chatting;

[0028] A behavior level evaluation unit, connected to the behavior recognition unit, is used to match the recognized action tags with the preset service quality level rules to generate corresponding service behavior level data;

[0029] The points control unit is connected to the behavior level evaluation unit and is used to adjust the points assignment weight corresponding to the service duration data according to the service behavior level data, and store the adjusted points data in the dynamic service data.

[0030] In an exemplary embodiment of the present disclosure, the feature extraction module includes:

[0031] A data access unit, which establishes a data interaction channel with the volunteer information database and the service quality analysis module, and is used to obtain static identity information, historical service record data, and service behavior level data corresponding to candidate volunteers;

[0032] a service data integration unit connected to the data access unit and the service duration recording module, configured to perform field alignment and format normalization processing on the historical service record data, the service behavior level data, and the service duration data to generate integrated service data;

[0033] a feature quantification unit connected to the service data integration unit, configured to convert parameters of multiple dimensions such as service frequency, service area coverage, service behavior level, and user rating in the integrated service data into corresponding numerical feature items and perform normalization processing;

[0034] A feature vector generating unit is connected to the feature quantizing unit and is used to construct a feature vector of a standard structure based on the numerical feature item and provide the feature vector to the service matching module.

[0035] In an exemplary embodiment of the present disclosure, the service matching module includes:

[0036] A model access unit, connected to the feature extraction module, for calling preset machine learning model parameters and inputting the feature vector into the machine learning model;

[0037] a matching score calculation unit, connected to the model access unit, configured to output corresponding matching score data based on a calculation result of the feature vector in the machine learning model;

[0038] A score sorting unit, connected to the matching score calculation unit, is used to sort the matching score data according to a preset priority rule to generate a candidate volunteer ranking result;

[0039] A matching data generating unit is connected to the score ranking unit and is used to extract matching priority items based on the ranking results and construct corresponding matching relationship data.

[0040] According to a second aspect of the present disclosure, a time bank volunteer scheduling method for smart elderly care services is provided, comprising:

[0041] Presetting coordinate parameters of an electronic geo-fence of a target service area, and generating corresponding entry and exit event data when a mobile terminal enters or leaves the electronic geo-fence;

[0042] Determine the service start time and service end time of each volunteer based on the entry and exit event data, and generate corresponding service duration data accordingly;

[0043] Used to obtain static information and dynamic service data related to candidate volunteers from the database, and perform feature extraction processing on the information in combination with the service duration data to generate a feature vector for matching;

[0044] Based on the feature vector and the preset machine learning model, the matching score between the candidate volunteer and the target service object is calculated, and matching relationship data is generated based on the matching score.

[0045] According to a third aspect of the present disclosure, a computer-readable storage medium is provided, on which a computer program is stored. When the computer program is executed by a processor, any step of the method described in the second aspect is implemented.

[0046] According to a fourth aspect of the present disclosure, there is provided an electronic device, including:

[0047] A processor and a memory, wherein the memory stores computer-readable instructions, and when the computer-readable instructions are executed by the processor, any step of the method described in the second aspect is implemented.

[0048] The technical solutions provided by the embodiments of the present disclosure may have the following beneficial effects:

[0049] In the time banking system of the exemplary embodiments disclosed herein, the geofence detection module is based on a mechanism that compares positioning signals with coordinate parameters. It doesn't rely on user interaction to generate "entry and exit" events. Instead, it uses objective spatial state changes to construct service boundaries, ensuring that service behaviors are subject to "dual constraints of space and time," thus naturally mitigating the potential for time tampering or false triggering caused by human intervention. The event data generated by the module clearly identifies the entry and exit boundary nodes of a space, forming the temporal reference starting and ending points for subsequent service behaviors, allowing the system to process service data based on unified physical triggering conditions. The service duration recording module, based on the identification of entry and exit events, does not perform secondary subjective processing on the raw time data. Instead, it calculates service time intervals based on entry and exit events, ensuring that the generated service duration data has a clear structure and traceable path. The service matching module uses standard feature vectors as input and invokes a machine learning model for scoring and ranking. Matching results are directly generated by the system, eliminating manual coordination or screening and maintaining a closed-loop and repeatable matching decision process. In summary, the present disclosure can complete a complete set of technical processes of "service verification - information structuring - model judgment" without human intervention, and complete the establishment of a closed loop of time bank services in smart elderly care services.

[0050] It is to be understood that the foregoing general description and the following detailed description are exemplary and explanatory only and are not restrictive of the disclosure. BRIEF DESCRIPTION OF THE DRAWINGS

[0051] The accompanying drawings are incorporated into and constitute a part of the specification, illustrate embodiments consistent with the present disclosure, and together with the specification, are used to explain the principles of the present disclosure. Obviously, the drawings described below are only some embodiments of the present disclosure, and those skilled in the art can derive other drawings based on these drawings without inventive effort.

[0052] Figure 1 A system architecture diagram of a time bank volunteer scheduling method for smart elderly care services to which an embodiment of the present disclosure can be applied is shown.

[0053] Figure 2 A schematic module block diagram of a time bank system for intelligent elderly care services in an embodiment of the present disclosure is shown.

[0054] Figure 3 A schematic module block diagram of another time bank system for smart elderly care services in an embodiment of the present disclosure is shown.

[0055] Figure 4 A schematic module block diagram of another time bank system for smart elderly care services in an embodiment of the present disclosure is shown.

[0056] Figure 5 A flow chart of a method for scheduling time bank volunteers for smart elderly care services in an embodiment of the present disclosure is shown.

[0057] Figure 6 A schematic structural diagram of an electronic device suitable for implementing the embodiments of the present disclosure is shown.

[0058] In the drawings, the same or corresponding reference numerals denote the same or corresponding parts. DETAILED DESCRIPTION

[0059] The terms used in this specification are for the purpose of describing specific embodiments only and are not intended to limit this specification. As used in this specification and the appended claims, the singular forms "a," "an," "the," and "the" are intended to include the plural forms as well, unless the context clearly indicates otherwise. It should also be understood that the term "and / or" as used herein refers to and encompasses any and all possible combinations of one or more of the associated listed items.

[0060] It should be understood that although the terms first, second, third, etc. may be used in this specification to describe various information, such information should not be limited to these terms. These terms are merely used to distinguish information of the same type from one another. For example, first information may also be referred to as second information, and similarly, second information may also be referred to as first information without departing from the scope of this specification. Depending on the context, the term "if" as used herein may be interpreted as "when," "when," or "in response to determining."

[0061] Figure 1 A schematic diagram of the system architecture of a time bank volunteer scheduling method for intelligent elderly care services to which an embodiment of the present disclosure can be applied is shown.

[0062] like Figure 1 As shown, system architecture 100 may include one or more terminal devices such as a smartphone 101, a tablet computer 102, a portable computer 103, a network 104, and a server 105. Network 104 is a medium for providing a communication link between the terminal device and server 105. Network 104 may include various connection types, such as wired or wireless communication links or fiber optic cables.

[0063] The terminal device can be any electronic device with data processing capabilities, which has a display screen for displaying service recipient information, time bank information, etc. to the user. The electronic device includes but is not limited to the above-mentioned portable computers, smart phones, tablet computers, etc.

[0064] It should be understood that Figure 1 The number of terminal devices, networks and servers in the embodiment is merely illustrative. Any number of terminal devices, networks and servers may be provided as needed. For example, the server 105 may be a server cluster consisting of multiple servers.

[0065] The time bank volunteer scheduling method for smart elderly care services provided in the embodiment of the present disclosure can be executed by a terminal device, and accordingly, the device for displaying product images on a mobile device in a low-bandwidth environment can be set in the terminal device. However, it is easy for those skilled in the art to understand that the time bank volunteer scheduling method for smart elderly care services provided in the embodiment of the present disclosure can also be executed by the server 105, and accordingly, the device for displaying product images on a mobile device in a low-bandwidth environment can also be set in the server 105, and this exemplary embodiment does not specifically limit this.

[0066] The present disclosure provides a time bank system for intelligent elderly care services. Figure 2As shown, the time bank system 200 for smart elderly care services may include an electronic geo-fence detection module 210 , a service duration recording module 220 , a feature extraction module 230 and a service matching module 240 .

[0067] in:

[0068] The electronic geofence detection module 210 is configured to pre-set the coordinate parameters of the electronic geofence of the target service area and generate corresponding entry and exit event data when the mobile terminal enters or leaves the electronic geofence.

[0069] refer to Figure 3 As shown, in this exemplary embodiment, the electronic geo-fence detection module 210 includes a coordinate parameter receiving unit 2101, a location information collecting unit 2102, a fence determination unit 2103, and an event generating unit 2104.

[0070] The coordinate parameter receiving unit 2101 is configured to load the coordinate parameters of the electronic geographic fence corresponding to the target service area from a system preset configuration file and store the coordinate parameters in a local buffer.

[0071] In this example implementation, the coordinate parameters are described in a polygonal boundary description format encoded in a unified format, and the structure can be parsed in GeoJSON or WKT (Well-Known Text) format. In this embodiment, the fence boundary can be composed of at least three vertex coordinates to form a two-dimensional closed area, and the single point coordinate expression is P i =(lat i ,lon i ), where lat is latitude and lon is longitude. In this example embodiment, the system can preset fence parameters with different levels of accuracy based on the geographical characteristics of the service location, preferably with an accuracy within ±5 meters, and set multiple sets of fences in the configuration file for switching based on the application scenario.

[0072] Furthermore, the receiving unit stores the loaded coordinate parameters in a local buffer to reduce repeated network requests and ensure operational stability in the event of a disconnection. The buffer can be deployed in the terminal's built-in storage (such as eMMC or UFS) or RAM space, and a timeout period T can be set accordingly. cache , for example, no repeated loading within 10 minutes.

[0073] The position information collecting unit 2102 is in communication with the coordinate parameter receiving unit 2101 and is used to collect the current position information in real time through the positioning component of the mobile terminal and generate position information data for fence determination.

[0074] In the example embodiment, the above acquisition process adopts a multi-source fusion positioning strategy, including the cooperation of a global satellite positioning system (GNSS) module and a wireless local area network scanning module. The GNSS positioning can call the GPS, GLONASS, BeiDou or Galileo positioning components integrated in the mobile terminal chip to obtain the positioning result, and the default sampling frequency is f GNSS = 1 Hz, and the output format is (lat g , lon g ). At the same time, the Wi-Fi location information acquisition module periodically scans the wireless access points (APs) that can be detected in the current environment, and constructs a signal fingerprint vector S w = [(id1, rssi1), …, (id n , rssi n )] based on the collected access point identifiers (BSSIDs), signal strengths (RSSIs), channel numbers and MAC addresses. The fingerprint vector is input to match with the local or cloud preset Wi-Fi signal fingerprint database, and the K nearest neighbor (KNN), Gaussian process regression or deep neural network inference model outputs the corresponding equivalent estimated location information (lat w , lon w ), wherein the estimation accuracy is represented by the prediction variance σ w 2 , and the confidence is determined by the corresponding prediction score or residual size.

[0075] The fusion operation of the above GNSS and Wi-Fi estimated location information can be realized by a weighted average model, and the fusion output location information is recorded as (lat f , lon f ), and the calculation formula is:

[0076] (lat f , lon f ) = α·(lat g , lon g ) + (1-α)·(lat w , lon w )

[0077] Wherein, α is a dynamic weight coefficient, satisfying α∈[0,1]. The value of α is determined according to the GNSS signal stability G sig and the Wi-Fi signal matching confidence C w , and the preferred weight adjustment function is:

[0078] α = softmax(κ g ·G sig , κ w ·C w )

[0079] where κ g is a scaling factor that controls the priority of GNSS and Wi-Fi in the fusion strategy. When GNSS signal strength is lower than a set threshold θ w ( e.g. θ g = -120 dBm) or the prediction residual σ g w 2 is lower than a tolerance upper limit ε w 2 ( e.g. ε w 2 = 225, corresponding to a maximum error of 15 meters), the system increases the weight of the Wi-Fi part to compensate for the situation where satellite signals are unavailable or jump failure. If the Wi-Fi matching accuracy is insufficient, the system will force α = 1, and only use GNSS results as the fusion output. The final obtained fusion position information (lat f , lon f ) meets the unified coordinate expression format and has a precision evaluation mechanism, serving as data input for subsequent electronic geofencing space judgment.

[0080] A fence judgment unit 2103 connected with the position information acquisition unit 2102 is configured to perform spatial comparison between the position information data and the coordinate parameters to generate a judgment mark indicating an entering state or a leaving state.

[0081] In the example embodiment, the comparison operation adopts a point-in-polygon judgment algorithm, which can adopt a Ray Casting method or a Winding Number method. Let the fusion positioning point be P f , and the fence polygon be Poly = [P1, P2,..., P n ]. If the algorithm judges that P f ∈ Poly, a state mark S = "IN" is generated, otherwise S = "OUT". The judgment is repeated and refreshed every Δt seconds, and Δt is usually set to 1-3 seconds. In addition, the device moving speed can be used for adaptive adjustment to reduce redundant calculation in the stationary state.

[0082] An event generation unit 2104 connected with the fence judgment unit 2103 is configured to construct corresponding entry and exit event data based on the judgment mark, and transmit the entry and exit event data to the service duration recording module 220.

[0083] In the example embodiment, the monitoring judgment mark state change process can be: when S changes from "OUT" to "IN", the entry event timestamp T in is recorded; and when S changes from "IN" to "OUT", the exit event timestamp T out is recorded., and construct the entry and exit event data E={T in ,T out ,UID}, where UID is the current mobile terminal user identifier. The event data structure can be encoded in JSON or Protobuf format to facilitate compressed transmission. The event generation unit 2104 uploads this event data to the service duration recording module 220 in real time or periodically. In some exemplary embodiments, an edge caching mechanism can also be employed. When network connectivity is abnormal, event data can be cached locally and uploaded in batches after the connection is restored.

[0084] The service duration recording module 220 is in communication with the electronic geo-fence detection module 210 and is used to determine the service start time and service end time of each volunteer based on the entry and exit event data, and generate corresponding service duration data accordingly.

[0085] refer to Figure 3 As shown, in this example implementation, the service duration recording module 220 may include a timestamp extraction unit 2201 , a time interval calculation unit 2202 , and a duration generation unit 2203 .

[0086] in:

[0087] The timestamp extraction unit 2201 is in communication with the electronic geofence detection module 210 and is used to extract a first timestamp corresponding to the volunteer entering the electronic geofence and a second timestamp corresponding to the volunteer leaving the electronic geofence from the entry and exit event data.

[0088] In this example implementation, the timestamp extraction unit 2201 extracts key time nodes from the entry and exit event data. The entry and exit event data is a structured data packet constructed by the event generation unit 2104, and the format is preferably E event ={UID,S,T event}, where UID is the user ID and S is the event status (e.g. "S in "Indicates entry," S out ” means leaving), T event The timestamp extraction unit 2201 can sort the event streams with the same UID by time and use the “S in ” and “S out "The state switching logic extracts the first timestamp T corresponding to the entry event in The second timestamp T corresponding to the leaving event out , to form a matching time pair P i =(T in i ,T out i). In addition, if there is an entry but no exit event in the data stream, the system can set the maximum allowed service time T max (such as T max = 8 hours), and then automatically change the current system time or T in +T max Treated as T out The derived value of the exception is added to the record.

[0089] The time interval calculation unit 2202 is connected to the timestamp extraction unit 2201 and is configured to calculate service time interval data based on the difference between the first timestamp and the second timestamp.

[0090] In this example implementation, the time interval calculation unit 2202 receives the timestamp pair (T in ,T out ) After that, the service time interval ΔT is generated based on the time difference calculation logic, that is, ΔT = T 出 -T 入 In addition, in the cross-day or long-cycle service scenario, the time interval calculation unit 2202 in this example embodiment can further determine whether it crosses the day boundary or has an abnormal negative value. If ΔT≤0, it is automatically determined as invalid data and thrown into the exception queue Q err , waiting for subsequent log playback or manual review. The service time interval ΔT is in seconds and is structured as T range ={UID,ΔT,valid_flag}, where valid_flag is used to mark whether the interval data is legal and can be adopted by subsequent modules.

[0091] The duration generating unit 2203 is connected to the time interval calculating unit 2202 , and is configured to convert the service time interval data into service duration data in a standard format, and transmit the service duration data to the feature extracting module 230 .

[0092] In this example embodiment, in order to improve the modeling consistency of the feature extraction module 230, ΔT can be converted into the service time data D in a standard format. time Specific conversion methods can be linear normalization, interval discretization or integer minute mapping strategies; this exemplary embodiment does not make special restrictions on this. Preferably, this embodiment adopts the service unit time U t (e.g. 15 minutes) as the minimum unit of measurement, perform the following standard conversion:

[0093] T norm =round(ΔT / U t )×U t

[0094] After conversion, the Tnorm The final standard service duration value is formed and encapsulated together with UID, timestamp and other fields as follows:

[0095] D time ={UID,T in ,T out ,ΔT,T norm ,valid_flag}

[0096] The valid_flag field records whether the service time record has been completed and whether it meets the minimum service threshold ΔT min (e.g. 30 minutes), and attach a service exception label for reference by the subsequent feature modeling module. time The data is transmitted to the feature extraction module 230 in the form of structured data. The transmission method is preferably JSON format through RESTful API or Protocol Buffers via gRPC protocol to ensure low-latency and high-compatibility data docking.

[0097] The feature extraction module 230 is connected to the service time recording module 220 and establishes a data interaction channel with a preset volunteer information database, so as to obtain static information and dynamic service data related to the candidate from the database, and perform feature extraction processing on the information in combination with the service time data to generate a feature vector for matching.

[0098] In this example implementation, the dynamic service data includes service quality data. Figure 4 As shown, the time bank system may further include a service quality analysis module 250. The service quality analysis module 250 may include an image acquisition unit 2501, a behavior recognition unit 2502, a behavior level evaluation unit 2503, and a score control unit 2504.

[0099] The image acquisition unit 2501 is connected to the camera module of the mobile terminal and is used to periodically acquire video frame image data of the service site during the volunteer service process.

[0100] In this example implementation, the image acquisition unit 2501 directly calls the front or rear camera module integrated in the mobile terminal, and the sampling frame rate can be set to f cap = 1–3fps to balance the action granularity required for image recognition and the pressure on terminal computing resources. t By timestamp T t Encode and cache in local temporary frame buffer B img ,The frame cache strategy adopts a ring overlay structure to control the storage occupancy.

[0101] The behavior recognition unit 2502 is connected to the image acquisition unit 2501 and is used to analyze the human motion features in the video frame image data based on a preset motion recognition model, and identify action labels including one or more service behaviors such as raising hands, feeding, supporting, turning over, going to the toilet, cleaning, and chatting.

[0102] In this example embodiment, the action recognition model is preferably a time series recognition model jointly constructed based on a convolutional neural network (CNN) and a long short-term memory network (LSTM). The action recognition model can adopt a two-stage structure, wherein the first stage is to extract key points of the human skeleton, and the second stage is to perform action recognition classification on the key point time series information. Among them, the human body key point extraction can be implemented through the OpenPose or MediaPipe model, and the key point set of each frame is output:

[0103] K t ={(x i ,y i ,c i )} i＝1～n

[0104] 其中xi、yi为第i个关键点的坐标位置，ci为置信度系数。时序特征输入模型后，结合上下帧位置变化形成动作轨迹向量Fseq，模型输出对应动作类别标签La∈{抬手、喂饭、搀扶、翻身、如厕、清洁、聊天}。

[0105] 行为等级评估单元2503，其与所述行为识别单元2502连接，用于根据所识别的动作标签在预设服务质量等级规则中进行匹配，生成对应的服务行为等级数据。

[0106] 本示例实施方式中，行为等级评估单元2503以标签La为索引，查阅系统预设的服务行为等级映射表Rmap，执行规则匹配，输出等级值Gi∈{1,2,3,4,5}，其中数值越高表示服务强度越大、复杂度越高。映射规则可采用一一映射或多标签融合策略，例如：

[0107]

[0108]

[0109]

[0110] 评估单元在输出等级数据的同时，可记录每个识别周期内的主要动作组合及其触发时间，形成结构化服务行为等级数据Dgrade＝{UID,Tstart,Tend,La,Gi,valid_flag}，其中valid_flag指示该等级是否满足积分加权逻辑条件，如持续时间超过最小服务阈值、视频有效帧率达到识别精度要求等。

[0111] 积分调控单元2504，其与所述行为等级评估单元2503连接，用于根据所述服务行为等级数据调整所述服务时长数据对应的积分赋值权重，并将调整后的积分数据存储至所述动态服务数据。

[0112] 本示例实施方式中，积分调控单元2504接收Dgrade并与对应的服务时长数据Dtime联动处理；其中，原始积分值Praw由服务时长标准转换后生成，积分调控单元2504引入服务等级系数Wgrade，按照如下公式计算调控后积分值Padj：

[0113] Padj＝Praw×f(Gi)

[0114] 其中f(Gi)为权重函数，例如线性函数f(G)＝1+0.2×(G-1)，若G＝5则积分提高至原始积分的1.8倍。在本公开的其他示例性实施例中，也可以接入后台规则引擎以根据服务对象状态、时间段或服务紧急性动态调整f(G)。最终积分数据Dscore＝{UID,Tin,Tout,Gi,Padj}被存储至动态服务数据集中，供后续特征提取与历史评分系统引用。

[0115] 本示例实施方式中，所述特征提取模块230包括数据访问单元2301、服务数据整合单元2302、特征量化单元2303以及特征向量生成单元2304。其中：

[0116] 数据访问单元2301，与所述志愿者信息数据库和所述服务质量分析模块250建立有数据交互通道，用于获取与候选志愿者对应的静态信息与动态服务数据；所述静态信息包括身份信息，所述动态服务数据包括历史服务记录数据和服务行为等级数据。

[0117] 本示例实施方式中，所述静态信息包括但不限于身份信息字段，所述身份信息字段具体包括志愿者标识UID、年龄、性别、注册地区、联系方式及服务偏好标签等，用于表达志愿者的基本属性和背景特征；动态服务数据则包括历史服务记录数据与服务行为等级数据。其中，所述历史服务记录数据来源于服务时长记录模块220与数据库中的服务履历表，具体包括每一项服务对应的服务起始时间Tin、服务结束时间Tout、服务持续时长ΔT、服务对象类别C_target、积分记录P_hist、服务评分R_eval等字段，反映志愿者在过去服务周期内的服务数量、服务类型与用户反馈情况。所述服务行为等级数据已由服务质量分析模块250生成，此处不再赘述。

[0118] 服务数据整合单元2302，与所述数据访问单元2301以及所述服务时长记录模块220连接，用于将所述历史服务记录数据、所述服务行为等级数据与所述服务时长数据进行字段对齐与格式归一化处理，生成整合服务数据；

[0119] 本示例实施方式中，所述服务数据整合单元2302以UID为关联索引，将Dgrade中的Gi值与对应时段内的Dtime进行绑定，同时引用Sstatic提取静态字段，形成结构化服务数据记录Fraw。字段对齐操作中，时间字段统一转化为ISO8601格式，区域字段region与服务地点信息统一映射为行政区代码Acode，动作标签La可以仅保留用于审核和解释，不进入特征建模。服务频次通过统计近N天内的Dtime数量计算得出，服务区域覆盖率通过计算近N次服务涉及的唯一Acode数量与可服务区域总数之比得出，服务等级取近N次Dgrade中Gi的加权平均值，用户评分来自H_service中R_eval字段。

[0120] 整合后的服务特征结构定义例如可以为：Fraw:＝{F1,F2,F3,F4}。其中，F1表示服务频次；F2表示服务区域覆盖率(范围：[0,1])；F3表示服务行为等级均值(来源Gi)；F4表示用户评分加权平均值。

[0121] 特征量化单元2303，与所述服务数据整合单元2302连接，用于将所述整合服务数据中的服务频次、服务区域覆盖率、服务行为等级、用户评分中的多个维度的参数转换为对应的数值型特征项，并进行归一化处理。

[0122] 本示例实施方式中，特征量化单元2303将上述Fraw中的所有参数转换为标准化数值型特征项Fnum:＝{x1,x2,x3,x4}。归一化方法可以为最小-最大归一化，其表达式为：xi′＝(xi-mini) / (maxi-mini)。其中xi′表示归一化后数值，mini和maxi分别表示第i项参数在系统样本库中统计得到的最小值和最大值。例如，若服务频次F1＝16，统计范围为[0,60]，则归一化结果为x1′＝(16-0) / (60-0)＝0.267。

[0123] 特征向量生成单元2304，与所述特征量化单元2303连接，用于基于所述数值型特征项构建标准结构的特征向量，并将所述特征向量提供至所述服务匹配模块240。

[0124] 本示例实施方式中，特征向量生成单元2304接收Fnum后生成特征向量V_feature：

[0125] V_feature:＝[x1′,x2′,x3′,x4′]

[0126] 本示例实施方式中，特征向量结构固定，维度为4，格式可采用JSON、NumPy array或Tensor表示，用于后续服务匹配模块240的机器学习模型调用。但本示例性实施例中并不以此为限。

[0127] 服务匹配模块240，与所述特征提取模块230连接，用于基于所述特征向量及预设的机器学习模型，计算候选志愿者与目标服务对象之间的匹配分值，并基于所述匹配分值生成匹配关系数据。

[0128] 本示例实施方式中，所述服务匹配模块240包括模型接入单元2401、匹配分值计算单元2402、分值排序单元2403以及匹配数据生成单元2404。其中：

[0129] 模型接入单元2401，与所述特征提取模块230连接，用于调用预设的机器学习模型参数，并将所述特征向量输入至所述机器学习模型中。

[0130] 本示例实施方式中，模型接入单元2401将从特征提取模块230接收到的特征向量V_feature输入至所述模型中进行计算处理。所述模型优选采用支持向量机(SVM)、梯度提升树(GBDT)、多层感知机(MLP)或图神经网络(GNN)等类型，可通过TensorFlow、PyTorch或scikit-learn框架进行训练与部署。模型输入结构与V_feature保持一致，其维度设定为n，表示包含n个数值型特征项，例如

[0131] 匹配分值计算单元2402，其与所述模型接入单元2401连接，用于基于所述特征向量在所述机器学习模型中的运算结果，输出对应的匹配分值数据。

[0132] 本示例实施方式中，以多层感知机模型为例，输入向量V_feature＝[x1,x2,…,xn]经由若干隐层加权运算与激活函数处理后，输出匹配分值S_match，其数学表达为：

[0133] h1＝ReLU(W1·V_feature+b1)

[0134] h2＝ReLU(W2·h1+b2)

[0135] S_match＝σ(Wout·h2+bout)

[0136] 其中W1、W2、Wout分别为各层权重矩阵，b1、b2、bout为对应偏置向量，ReLU表示线性整流激活函数，σ表示Sigmoid函数或Softmax函数，具体函数类型根据模型目标设定为二分类匹配评分或多类匹配概率。输出结果S_match∈[0,1]表示候选志愿者与当前服务对象的匹配概率或匹配程度，数值越高表示匹配优先级越高。对于一组候选者{V_feature1,V_feature2,…,V_featurem}，匹配分值计算单元2402将依次输出对应分值集合{S1,S2,…,Sm}，每个分值与一个特征向量保持一一对应关系。

[0137] 分值排序单元2403，其与所述匹配分值计算单元2402连接，用于对所述匹配分值数据按照预设优先级规则进行排序处理，生成候选志愿者排序结果。

[0138] 本示例实施方式中，排序依据为系统设定的优先级规则R_sort，该规则可基于分值大小进行降序排列，亦可根据加权策略引入地理位置距离D_geo、志愿者可用时间窗口T_avail与服务对象偏好匹配度P_pref等附加因子进行复合排序。若采用线性加权排序策略，排序因子定义为：

[0139] 总分i＝α×Si+β×(1-D_geoi / D_max)+γ×P_prefi

[0140] 其中α、β、γ为加权系数，满足α+β+γ＝1，D_max为系统设定的最大距离阈值，Score_totali为候选志愿者i的综合排序得分。排序结果输出为Rank_list:＝{UID1,UID2,…,UIDm}，其中UID1表示匹配优先级最高者，UIDm为最低者。

[0141] 匹配数据生成单元2404，其与所述分值排序单元2403连接，用于基于所述排序结果提取匹配优先项，并构造对应的匹配关系数据。

[0142] 本示例实施方式中，匹配数据生成单元2404可以提取Rank_list中排名靠前的若干个候选志愿者信息(可配置为Top-k)，并结合服务任务编号Task-ID、匹配分值S_match与原始特征向量索引，生成匹配数据结构M_result，定义为：

[0143] M_result:＝{Task-ID,[UIDi,Si,排名i]|i∈[1,k]}

[0144] 该结构可用于前端显示、任务指派、后台审核等业务模块调用，并可持久化存储至匹配结果数据库或缓存中以备查询。此外，在本公开的更多示例性实施例中，匹配数据生成单元2404还可以使用哈希函数Hash(Task-ID+UID+timestamp)为每一匹配关系生成唯一标识符Match-ID，保证匹配记录的可验证性与一致性。

[0145] 进一步的，本公开实施例还提供了一种用于智能养老服务的时间银行志愿者调度方法，参考图5所示，该方法可以包括步骤S510至步骤S540：

[0146] 步骤S510、预先设定目标服务区域的电子地理围栏的坐标参数，并在移动终端进入或离开所述电子地理围栏时，生成对应的进出事件数据；

[0147] 步骤S520、基于所述进出事件数据确定各愿者服务开始时间与服务结束时间，并据此生成对应的服务时长数据；

[0148] 步骤S530、用于从数据库中获取与候选志愿者相关的静态信息与动态服务数据，并结合所述服务时长数据对所述信息进行特征提取处理，生成用于匹配的特征向量；

[0149] 步骤S540、基于所述特征向量及预设的机器学习模型，计算候选志愿者与目标服务对象之间的匹配分值，并基于所述匹配分值生成匹配关系数据。

[0150] 上述中时间银行志愿者调度方法各步骤的具体细节已经在对应的时间银行系统中进行了详细的描述，因此此处不再赘述。

[0151] 应当注意，尽管在上文详细描述中提及了时间银行系统的若干模块或者单元，但是这种划分并非强制性的。实际上，根据本公开的实施方式，上文描述的两个或更多模块或者单元的特征和功能可以在一个模块或者单元中具体化。反之，上文描述的一个模块或者单元的特征和功能可以进一步划分为由多个模块或者单元来具体化。

[0152] 本公开的示例性实施方式还提供了一种计算机可读存储介质，其上存储有能够实现本说明书上述方法的程序产品。在一些可能的实施方式中，本公开的各个方面还可以实现为一种程序产品的形式，其包括程序代码，当程序产品在电子设备上运行时，程序代码用于使电子设备执行本说明书上述"示例性方法”部分中描述的根据本公开各种示例性实施方式的步骤。

[0153] 该程序产品可以采用便携式紧凑盘只读存储器(CD-ROM)并包括程序代码，并可以在电子设备，例如个人电脑上运行。然而，本公开的程序产品不限于此，在本文件中，可读存储介质可以是任何包含或存储程序的有形介质，该程序可以被指令执行系统、装置或者器件使用或者与其结合使用。

[0154] 程序产品可以采用一个或多个可读介质的任意组合。可读介质可以是可读信号介质或者可读存储介质。可读存储介质例如可以为但不限于电、磁、光、电磁、红外线、或半导体的系统、装置或器件，或者任意以上的组合。可读存储介质的更具体的例子(非穷举的列表)包括：具有一个或多个导线的电连接、便携式盘、硬盘、随机存取存储器(RAM)、只读存储器(ROM)、可擦式可编程只读存储器(EPROM或闪存)、光纤、便携式紧凑盘只读存储器(CD-ROM)、光存储器件、磁存储器件、或者上述的任意合适的组合。

[0155] 计算机可读信号介质可以包括在基带中或者作为载波一部分传播的数据信号，其中承载了可读程序代码。这种传播的数据信号可以采用多种形式，包括但不限于电磁信号、光信号或上述的任意合适的组合。可读信号介质还可以是可读存储介质以外的任何可读介质，该可读介质可以发送、传播或者传输用于由指令执行系统、装置或者器件使用或者与其结合使用的程序。

[0156] 可读介质上包含的程序代码可以用任何适当的介质传输，包括但不限于无线、有线、光缆、RF等等，或者上述的任意合适的组合。

[0157] 可以以一种或多种程序设计语言的任意组合来编写用于执行本公开操作的程序代码，程序设计语言包括面向对象的程序设计语言—诸如Java、C#、C++等，还包括常规的过程式程序设计语言—诸如"C”语言或类似的程序设计语言。程序代码可以完全地在用户计算设备上执行、部分地在用户设备上执行、作为一个独立的软件包执行、部分在用户计算设备上部分在远程计算设备上执行、或者完全在远程计算设备或服务器上执行。在涉及远程计算设备的情形中，远程计算设备可以通过任意种类的网络，包括局域网(LAN)或广域网(WAN)，连接到用户计算设备，或者，可以连接到外部计算设备(例如利用因特网服务提供商来通过因特网连接)。

[0158] 此外，本公开的示例性实施例还提供了一种能够实现上述用于智能养老服务的时间银行志愿者调度方法的电子设备。

[0159] 下面参照图6来描述根据本公开的这种实施例的电子设备600。图6所示的电子设备600仅仅是一个示例，不应对本公开实施例的功能和使用范围带来任何限制。

[0160] 如图6所示，电子设备600以通用计算设备的形式表现。电子设备600的组件可以包括但不限于：上述至少一个处理单元610、上述至少一个存储单元620、连接不同系统组件(包括存储单元620和处理单元610)的总线630、显示单元640。

[0161] 存储单元620存储有程序代码，程序代码可以被处理单元610执行，使得处理单元610执行本说明书上述"示例性方法”部分中描述的根据本公开各种示例性实施方式的步骤。例如，处理单元610可以执行本公开示例实施方式中的方法步骤。

[0162] 存储单元620可以包括易失性存储单元形式的可读介质，例如随机存取存储单元(RAM)621和 / 或高速缓存存储单元(Cache)622，还可以进一步包括只读存储单元(ROM)623。

[0163] 存储单元620还可以包括具有一组(至少一个)程序模块625的程序 / 实用工具624，这样的程序模块625包括但不限于：操作系统、一个或者多个应用程序、其它程序模块以及程序数据，这些示例中的每一个或某种组合中可能包括网络环境的实现。

[0164] 总线630可以为表示几类总线结构中的一种或多种，包括存储单元总线或者存储单元控制器、外围总线、图形加速端口、处理单元或者使用多种总线结构中的任意总线结构的局域总线。

[0165] 电子设备600也可以与一个或多个外部设备670(例如键盘、指向设备、蓝牙设备等)通信，还可与一个或者多个使得用户能与该电子设备600交互的设备通信，和 / 或与使得该电子设备600能与一个或多个其它计算设备进行通信的任何设备(例如路由器、调制解调器等等)通信。这种通信可以通过输入 / 输出(I / O)接口650进行。并且，电子设备600还可以通过网络适配器660与一个或者多个网络(例如局域网(LAN)，广域网(WAN)和 / 或公共网络，例如因特网)通信。如图所示，网络适配器660通过总线630与电子设备600的其它模块通信。应当明白，尽管图中未示出，可以结合电子设备600使用其它硬件和 / 或软件模块，包括但不限于：微代码、设备驱动器、冗余处理单元、外部磁盘驱动阵列、RAID系统、磁带驱动器以及数据备份存储系统等。

[0166] 通过以上的实施例的描述，本领域的技术人员易于理解，这里描述的示例实施例可以通过软件实现，也可以通过软件结合必要的硬件的方式来实现。因此，根据本公开实施例的技术方案可以以软件产品的形式体现出来，该软件产品可以存储在一个非易失性存储介质(可以是CD-ROM，U盘，移动硬盘等)中或网络上，包括若干指令以使得一台计算设备(可以是个人计算机、服务器、终端装置、或者网络设备等)执行根据本公开实施例的方法。

[0167] 此外，上述附图仅是根据本公开示例性实施例的方法所包括的处理的示意性说明，而不是限制目的。易于理解，上述附图所示的处理并不表明或限制这些处理的时间顺序。另外，也易于理解，这些处理可以是例如在多个模块中同步或异步执行的。

[0168] 通过以上的实施例的描述，本领域的技术人员易于理解，这里描述的示例实施例可以通过软件实现，也可以通过软件结合必要的硬件的方式来实现。因此，根据本公开实施例的技术方案可以以软件产品的形式体现出来，该软件产品可以存储在一个非易失性存储介质(可以是CD-ROM，U盘，移动硬盘等)中或网络上，包括若干指令以使得一台计算设备(可以是个人计算机、服务器、触控终端、或者网络设备等)执行根据本公开实施例的方法。

[0169] 本领域技术人员在考虑说明书及实践这里公开的发明后，将容易想到本公开的其它实施例。本申请旨在涵盖本公开的任何变型、用途或者适应性变化，这些变型、用途或者适应性变化遵循本公开的一般性原理并包括本公开未公开的本技术领域中的公知常识或惯用技术手段。说明书和实施例仅被视为示例性的，本公开的真正范围和精神由权利要求指出。

[0170] 应当理解的是，本公开并不局限于上面已经描述并在附图中示出的精确结构，并且可以在不脱离其范围进行各种修改和改变。本公开的范围仅由所附的权利要求来限制。

Claims

1. A time bank system for intelligent elderly care services, characterized in that: include: an electronic geofence detection module configured to pre-set coordinate parameters of an electronic geofence of a target service area and generate corresponding entry and exit event data when a mobile terminal enters or leaves the electronic geofence; a service duration recording module, in communication with the electronic geofence detection module, for determining the service start time and service end time of each volunteer based on the entry and exit event data, and generating corresponding service duration data accordingly; a feature extraction module connected to the service time recording module and establishing a data interaction channel with a preset volunteer information database, for obtaining static information and dynamic service data related to the candidate from the database, and performing feature extraction processing on the information in combination with the service time data to generate a feature vector for matching; The service matching module is connected to the feature extraction module and is used to calculate the matching score between the candidate volunteer and the target service object based on the feature vector and the preset machine learning model, and generate matching relationship data based on the matching score.

2. The time bank system for intelligent elderly care services according to claim 1 is characterized in that: The electronic geo-fence detection module includes: A coordinate parameter receiving unit is configured to load the coordinate parameters of the electronic geo-fence corresponding to the target service area from a system preset configuration file and store the coordinate parameters in a local buffer; a position information acquisition unit, communicatively connected to the coordinate parameter receiving unit, for acquiring current position information in real time through a positioning component of a mobile terminal, and generating position information data for fence determination; a fence determination unit connected to the position information acquisition unit and configured to perform a spatial comparison between the position information data and the coordinate parameters to generate a determination mark indicating an entry state or an exit state; An event generating unit is connected to the fence determination unit and is used to construct corresponding entry and exit event data based on the determination identifier and transmit the entry and exit event data to the service duration recording module.

3. The time banking system according to claim 2, characterized in that: The location information collection unit generates location information data for fence determination in the following manner: Calling the global satellite positioning component of the mobile terminal to obtain current GPS location information data; Scan the Wi-Fi access point information in the current environment and identify the corresponding Wi-Fi location information data based on the preset hotspot characteristic parameters; Performing weighted fusion of the GPS location information data and the Wi-Fi location information data to generate fused location information data for fence determination; A spatial position comparison operation is performed based on the fused position information data and the coordinate parameters of the electronic geographic fence to generate the determination identifier.

4. The time bank system for intelligent elderly care services according to claim 1 is characterized in that: The service duration recording module includes: a timestamp extraction unit, in communication with the electronic geofence detection module, for extracting, from the entry and exit event data, a first timestamp corresponding to a volunteer entering the electronic geofence and a second timestamp corresponding to a volunteer leaving the electronic geofence; a time interval calculation unit, connected to the timestamp extraction unit, for calculating service time interval data based on a difference between the first timestamp and the second timestamp; The duration generating unit is connected to the time interval calculating unit and is used to convert the service time interval data into service duration data in a standard format and transmit the service duration data to the feature extraction module.

5. The time bank system for intelligent elderly care services according to claim 1 is characterized in that: The time banking system further includes a service quality analysis module; the service quality analysis module includes: An image acquisition unit connected to the camera module of the mobile terminal and configured to periodically acquire video frame image data of the service scene during the volunteer service process; a behavior recognition unit connected to the image acquisition unit, configured to analyze human motion features in the video frame image data based on a preset motion recognition model, and identify action labels including one or more service behaviors such as raising hands, feeding, supporting, turning over, using the toilet, cleaning, and chatting; A behavior level evaluation unit, connected to the behavior recognition unit, is used to match the recognized action tags with the preset service quality level rules to generate corresponding service behavior level data; The points control unit is connected to the behavior level evaluation unit and is used to adjust the points assignment weight corresponding to the service duration data according to the service behavior level data, and store the adjusted points data in the dynamic service data.

6. The time bank system for intelligent elderly care services according to claim 5 is characterized in that: The feature extraction module includes: A data access unit establishes a data interaction channel with the volunteer information database and the service quality analysis module, and is used to obtain static information and dynamic service data corresponding to candidate volunteers; the static information includes identity information, and the dynamic service data includes historical service record data and service behavior level data; a service data integration unit connected to the data access unit and the service duration recording module, configured to perform field alignment and format normalization processing on the historical service record data, the service behavior level data, and the service duration data to generate integrated service data; a feature quantification unit connected to the service data integration unit, configured to convert parameters of multiple dimensions such as service frequency, service area coverage, service behavior level, and user rating in the integrated service data into corresponding numerical feature items and perform normalization processing; A feature vector generating unit is connected to the feature quantizing unit and is used to construct a feature vector of a standard structure based on the numerical feature item and provide the feature vector to the service matching module.

7. The time bank system for intelligent elderly care services according to claim 1 is characterized in that: The service matching module includes: A model access unit, connected to the feature extraction module, for calling preset machine learning model parameters and inputting the feature vector into the machine learning model; a matching score calculation unit, connected to the model access unit, configured to output corresponding matching score data based on a calculation result of the feature vector in the machine learning model; A score sorting unit, connected to the matching score calculation unit, is used to sort the matching score data according to a preset priority rule to generate a candidate volunteer ranking result; A matching data generating unit is connected to the score ranking unit and is used to extract matching priority items based on the ranking results and construct corresponding matching relationship data.

8. A time bank volunteer scheduling method for intelligent elderly care services, characterized in that: include: Presetting coordinate parameters of an electronic geo-fence of a target service area, and generating corresponding entry and exit event data when a mobile terminal enters or leaves the electronic geo-fence; Determine the service start time and service end time of each volunteer based on the entry and exit event data, and generate corresponding service duration data accordingly; Used to obtain static information and dynamic service data related to candidate volunteers from the database, and perform feature extraction processing on the information in combination with the service duration data to generate a feature vector for matching; Based on the feature vector and the preset machine learning model, the matching score between the candidate volunteer and the target service object is calculated, and matching relationship data is generated based on the matching score.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the time bank volunteer scheduling method according to claim 8 is implemented.

10. An electronic device, characterized in that: include: processor; and A memory having computer-readable instructions stored thereon, wherein the computer-readable instructions, when executed by the processor, implement the time bank volunteer scheduling method according to claim 8.

Images