Personnel tumble alarm device and method based on wireless sensor network

Abstract

The invention relates to the field of personnel tumble alarm, and particularly discloses a personnel tumble alarm device and method based on a wireless sensor network. A time queue of real-time posture data of a user and static standing posture data are collected through an acceleration sensor module and an indoor positioning module, and the posture deviation degree between the real-time posture data and the static data of each user is calculated; carrying out context time sequence coding on each attitude deviation degree by adopting a data coding and processing technology based on artificial intelligence, carrying out embedded coding on each real-time attitude data, and then carrying out attitude jump degree time sequence transmission coding on each user real-time attitude embedded feature; therefore, according to the multi-dimensional representation between the posture change time sequence mode implicit features and the posture offset time sequence features after the codes are transmitted, an identification result is automatically generated, and a corresponding fall early warning prompt is generated.

Description

Technical Field

[0001] The present application relates to the field of personnel fall alarm, and specifically to a personnel fall alarm device and method based on a wireless sensor network. Background Art

[0002] In today's society, the aging population is deepening, and the number of elderly people living alone is increasing. As their physical functions decline, their balance ability and reaction speed decrease, falling has become one of the serious safety hazards they face. This not only brings great pain to themselves, but also increases the medical burden on families and society.

[0003] In this regard, the invention with publication number CN107909771A proposes a personnel fall alarm system based on a wireless sensor network and its implementation method, which monitors posture changes through an acceleration sensor worn by the user, and records the standing posture as a reference when starting. Calculate and filter the posture deviation, count the deviation probability and calculate the acceleration variance to distinguish between falls in violent exercise and static state. And use the indoor positioning module to confirm the position of the person. Once an accidental fall is confirmed, trigger a local alarm, take photos of the scene and push information.

[0004] This invention mainly relies on traditional threshold comparison, variance calculation and statistical analysis methods to judge falls. This method has limitations when dealing with nonlinear or complex posture change patterns, and it is difficult to capture and mine potential subtle changes, especially in environments with diverse user behaviors, and may not be able to effectively identify real fall events. In addition, this method requires manual setting of thresholds and other parameters, which may require recalibration for different users or environments, and lacks adaptive capabilities, which may lead to increased false alarm rates or missed alarms in certain situations.

[0005] Therefore, an optimized personnel fall alarm solution based on wireless sensor networks is desired. Summary of the invention

[0006] This application is made in consideration of the above problems. One object of this application is to provide a person fall alarm device and method based on a wireless sensor network.

[0007] The present application provides a method for alarming a person falling down based on a wireless sensor network, which comprises: S1: Wear the acceleration sensor module and place the indoor positioning module; S2: When starting, prompt the user to stand steadily and obtain static standing posture data; S3: collecting the user's real-time posture data to obtain a time queue of the user's real-time posture data, and calculating a time queue of posture deviation based on the static standing posture data; S4: performing time series encoding on the time queue of the posture deviation to obtain an implicit encoding vector of the posture deviation time series feature; S5: performing embedding coding on each user real-time posture data in the time queue of the user real-time posture data to obtain a time queue of user real-time posture embedding coding vectors; S6: performing temporal context coding of the gesture jump degree on the time queue of the user's real-time gesture embedded coding vector to obtain a gesture change temporal pattern implicit coding vector; S7: concatenating the posture change timing pattern implicit coding vector and the posture offset timing feature implicit coding vector to obtain a user posture timing multi-dimensional implicit coding vector; S8: Obtain a recognition result based on the multi-dimensional implicit coding vector of the user posture time series, and generate a fall warning prompt based on the recognition result.

[0008] For example, according to the personnel fall alarm method based on wireless sensor network of the present application, wherein, based on the static standing posture data, a time queue of posture deviation is calculated, including: calculating the posture deviation of the user's real-time posture data at each predetermined time point in the time queue of the user's real-time posture data relative to the static standing posture data to obtain the time queue of the posture deviation.

[0009] For example, according to the personnel fall alarm method based on wireless sensor network of the present application, wherein, the S4 includes: using a forward LSTM-based sequence encoder to perform time series encoding on the time queue of the posture deviation to obtain the implicit coding vector of the posture deviation time series feature.

[0010] For example, according to the personnel fall alarm method based on wireless sensor network of an embodiment of the present application, wherein the S5 includes: using a posture embedding encoder based on a fully connected layer to fully connect encode each user real-time posture data in the time queue of the user real-time posture data to obtain the time queue of the user real-time posture embedding coding vector.

[0011] For example, according to the personnel fall alarm method based on wireless sensor network of the present application, wherein said S6 includes: Calculate the message propagation significance factor of each user real-time posture embedding coding vector in the time queue of the user real-time posture embedding coding vector to obtain the time queue of the user real-time posture message propagation significance factor; Perform gated propagation on the time queue of the user's real-time posture message propagation significance factor to obtain the time queue of the user's real-time posture message propagation significance weight; The weighted sum of the time queues of the user's real-time posture embedded coding vectors is calculated based on the time queues of the user's real-time posture message propagation significance weights to obtain the implicit coding vector of the posture change timing pattern.

[0012] For example, according to the personnel fall alarm method based on wireless sensor network of the present application, the message propagation significance factor of each user real-time posture embedded coding vector in the time queue of the user real-time posture embedded coding vector is calculated to obtain the time queue of the user real-time posture message propagation significance factor, including: Calculate the feature jump degree of each user real-time posture embedding coding vector in the time queue of the user real-time posture embedding coding vector to obtain the time queue of the user real-time posture feature jump degree; Calculating the message propagation space span of each user real-time posture embedding coding vector in the time queue of the user real-time posture embedding coding vector to obtain the time queue of the user real-time posture message propagation space span; Based on the characteristic jump degree and message propagation space span of each user real-time posture embedding coding vector in the time queue of the user real-time posture embedding coding vector, the message propagation significance factor of each user real-time posture embedding coding vector is calculated to obtain the time queue of the user real-time posture message propagation significance factor.

[0013] For example, according to the personnel fall alarm method based on wireless sensor network of the present application, the time queue of the significant factor of the user's real-time posture message propagation is gated to obtain the time queue of the significant weight of the user's real-time posture message propagation, including: Performing a normalization process based on a softmax function on the time queue of the user's real-time posture message propagation significance factor to obtain a normalized time queue of the user's real-time posture message propagation significance factor; The time queue of the normalized user real-time posture message propagation significance factor is screened based on a mask function to obtain the time queue of the user real-time posture message propagation significance weight.

[0014] For example, according to the personnel fall alarm method based on wireless sensor network of the present application, wherein said S8 includes: Inputting the user posture temporal multi-dimensional implicit coding vector into a classifier-based state recognizer to obtain the recognition result; In response to the recognition result that the user has fallen, a fall warning prompt is generated.

[0015] The present application also provides a personnel fall alarm device based on a wireless sensor network, which includes: A data acquisition module, used to prompt the user to stand steadily and obtain static standing posture data when starting; A collection and calculation module, used for collecting the user's real-time posture data to obtain a time queue of the user's real-time posture data, and calculating a time queue of posture deviation based on the static standing posture data; A time series encoding module, used for performing time series encoding on the time queue of the posture deviation to obtain an implicit encoding vector of the posture deviation time series feature; An embedding coding module, used for embedding coding each user real-time posture data in the time queue of the user real-time posture data to obtain a time queue of user real-time posture embedding coding vectors; A temporal context coding module, used for performing temporal context coding of the gesture jump degree on the time queue of the user's real-time gesture embedding coding vector to obtain an implicit coding vector of the gesture change temporal pattern; A vector concatenation module, used for concatenating the posture change timing pattern implicit coding vector and the posture offset timing feature implicit coding vector to obtain a user posture timing multi-dimensional implicit coding vector; The fall warning prompt module is used to obtain a recognition result based on the multi-dimensional implicit coding vector of the user posture time sequence, and generate a fall warning prompt based on the recognition result.

[0016] For example, according to the wireless sensor network-based personnel fall alarm device of the present application, the collection and calculation module is used to: calculate the posture deviation of the user's real-time posture data at each predetermined time point in the time queue of the user's real-time posture data relative to the static standing posture data to obtain the time queue of the posture deviation.

[0017] According to the personnel fall alarm device and method based on wireless sensor network of the present application, it collects the time queue of user's real-time posture data and static standing posture data through the acceleration sensor module and the indoor positioning module, and calculates the posture deviation between each user's real-time posture data and the static data, and adopts artificial intelligence-based data encoding and processing technology to perform context time series encoding on each posture deviation, and embeds each real-time posture data at the same time, and then performs posture jump degree time series transfer encoding on each user's real-time posture embedded feature, so as to automatically generate recognition results according to the multi-dimensional representation between the implicit features of the posture change time series pattern after the transfer encoding and the posture deviation time series features, and generate corresponding fall warning prompts. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the following briefly introduces the drawings of the embodiments of the present application. Obviously, the drawings described below only relate to some embodiments of the present application, and are not intended to limit the present application.

[0019] Figure 1A flow chart of a method for alarming a person falling down based on a wireless sensor network in an embodiment of the present application is shown; Figure 2 A flowchart of sub-step S6 of the method for alarming a person falling down based on a wireless sensor network in an embodiment of the present application is shown; Figure 3 A flowchart of sub-step S61 of the method for alarming a person falling down based on a wireless sensor network in an embodiment of the present application is shown; Figure 4 A flowchart of sub-step S62 of the method for alarming a person falling down based on a wireless sensor network in an embodiment of the present application is shown; Figure 5 A schematic structural diagram of a personnel fall alarm device based on a wireless sensor network in an embodiment of the present application is shown. DETAILED DESCRIPTION

[0020] The terms used in this specification are those common terms currently widely used in the art in consideration of the functions of the present application, but these terms may vary according to the intentions of those skilled in the art, precedents, or new technologies in the art. In addition, specific terms may be selected, and in this case, their detailed meanings will be described in the detailed description of the present application. Therefore, the terms used in the specification should not be understood as simple names, but rather as a general description based on the meaning of the terms and the present application.

[0021] Although the present application makes various references to certain modules in the system according to the embodiments of the present application, any number of different modules can be used and run on the user terminal and / or server. The modules are only illustrative, and different aspects of the system and method can use different modules.

[0022] Flowcharts are used in the present application to illustrate the operations performed by the system according to the embodiments of the present application. It should be understood that the preceding or following operations are not necessarily performed accurately in order. On the contrary, various steps may be processed in reverse order or simultaneously as required. At the same time, other operations may also be added to these processes, or a certain step or several steps of operations may be removed from these processes.

[0023] The technical solutions in the embodiments of the present application will be described clearly and completely below in conjunction with the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present application, not all embodiments. Based on the embodiments of the present application, all other embodiments obtained by ordinary technicians in this field without creative work also fall within the scope of protection of the present application.

[0024] Based on this, the present application proposes a method for personnel fall alarm based on wireless sensor network, the technical concept of which is to collect the time queue and static standing posture data of the user's real-time posture data through the acceleration sensor module and the indoor positioning module, and calculate the posture deviation between the real-time posture data and the static data of each user, and use the data encoding and processing technology based on artificial intelligence to perform context time series encoding on each posture deviation, and embed the encoding on each real-time posture data at the same time, and then perform posture jump degree time series transfer encoding on the embedded features of each user's real-time posture, so as to automatically generate recognition results according to the multi-dimensional representation between the implicit features of the posture change time series pattern after the transfer encoding and the posture deviation time series features, and generate corresponding fall warning prompts. The present application can accurately capture the potential slight changes in posture data and efficiently identify nonlinear and complex posture patterns. In addition, it can automatically adapt to the behavioral habits of different users without frequent manual calibration of thresholds and other parameters, ensuring consistency and reliability in diverse environments.

[0025] Figure 1 FIG. 2 shows a flow chart of a method for alarming a person falling down based on a wireless sensor network in an embodiment of the present application. Figure 1 As shown, according to the embodiment of the present application, the method for alarming a person falling down based on a wireless sensor network includes the following steps: S1: wearing an acceleration sensor module and placing an indoor positioning module; S2: prompting the user to stand steadily and obtaining static standing posture data when starting; S3: collecting the user's real-time posture data to obtain a time queue of the user's real-time posture data, and calculating the time queue of posture deviation based on the static standing posture data; S4: performing time series encoding on the time queue of posture deviation to obtain an implicit encoding vector of posture offset time series features; S5: encoding the time queue of the user's real-time posture data The real-time posture data of each user are embedded and encoded to obtain a time queue of the user's real-time posture embedded coding vector; S6: the time queue of the user's real-time posture embedded coding vector is subjected to temporal context coding of the posture jump degree to obtain an implicit coding vector of the posture change timing pattern; S7: the implicit coding vector of the posture change timing pattern and the implicit coding vector of the posture offset timing feature are concatenated to obtain a user posture timing multi-dimensional implicit coding vector; S8: based on the user posture timing multi-dimensional implicit coding vector, a recognition result is obtained, and based on the recognition result, a fall warning prompt is generated.

[0026] Specifically, in the technical solution of the present application, first, an acceleration sensor module is worn and an indoor positioning module is placed.

[0027] Next, when starting up, the user is prompted to stand steadily and obtain static standing posture data. (The system prompts the user to stand still for a period of time (for example, 5 seconds). During this period, the acceleration sensor collects the components of gravity acceleration on the X, Y, and Z axes at a high frequency (such as 50Hz). The collected data will be used to calculate the average value to obtain the distribution of gravity acceleration in the three-dimensional coordinate system in the current state. This set of data is regarded as "static standing data", which represents the body posture characteristics of the user when standing normally.) Then, the user's real-time posture data is collected to obtain a time queue of the user's real-time posture data. Specifically, in the personnel fall alarm method based on wireless sensor network, data collection can be completed by an acceleration sensor module. The module is worn at the waist of the person. When the system is started and the initial static standing posture recognition is completed, it enters the real-time monitoring stage. In this stage, the acceleration sensor continuously collects the wearer's three-axis acceleration information at a frequency of 50Hz. This information reflects the distribution of gravity acceleration in the acceleration sensor's own coordinate system and changes with the wearer's body posture. The acceleration value obtained by each sampling will be decomposed into specific values ​​in the three dimensions of X, Y, and Z, respectively expressed as X(i), Y(i), and Z(i), where i represents the index of the i-th sampling. These data will then be sent to the local server for processing by the acceleration sensor module through the ZigBee protocol. As the data continues to flow in, the local server stores the acceleration components received at each moment to form a time queue containing the user's real-time posture data. In order to ensure the validity and accuracy of the data, the server will also perform mean filtering on the posture deviation in the recent period to reduce the slight jitter caused by background noise interference. To this end, the server uses a circular queue array H with a length of 150 to save the filtered posture deviation data in the past 3 seconds. Every time a new sample is received, the array will be updated, and the oldest data will be replaced with the latest data, so as to ensure that all elements in the array are the latest measurement results within 3 seconds. In this way, through continuous real-time posture data collection and processing, the system can construct a time series that accurately reflects the user's posture changes, laying the foundation for the subsequent calculation of the implicit coding vector of the posture offset timing feature and the implicit coding vector of the posture change timing pattern, thereby realizing the timely detection and early warning of potential fall events.

[0028] Next, it is considered that the static standing posture data represents the posture of the user in a normal static state, while the user's real-time posture data reflects the posture in actual activities. Since the human body's posture will constantly change during activities, from standing to walking, turning around and other actions. Therefore, in the technical solution of the present application, based on the static standing posture data, a time queue of posture deviation is calculated, including: calculating the posture deviation of the user's real-time posture data at each predetermined time point in the time queue of the user's real-time posture data relative to the static standing posture data to obtain the time queue of the posture deviation. In other words, the posture deviation quantifies the difference between the user's posture and the benchmark at each time point. In this way, these changes in different user behavior patterns can be effectively captured, especially subtle but continuous changes, which may be early signals of abnormal situations such as falls.

[0029] It should be understood that the process of calculating the posture deviation begins with obtaining static standing posture data, which is collected within the first 5 seconds after the user starts the accelerometer module, during which the system prompts the user to stand still and remain stable. After processing the 250 acceleration data in these 5 seconds (because the sampling frequency is 50Hz), the average acceleration value on each axis is generated, and these average values ​​represent the user's posture baseline when standing still. Once the static standing posture data is used as a reference, the system can start to monitor the posture changes of the user in real time. During the real-time monitoring stage, the accelerometer sends new three-axis acceleration data to the local server at a frequency of 50Hz every second. For each real-time acceleration data collected at a time point, the system compares it with the average value of static standing to evaluate the change of the current posture. Specifically, the system compares the difference between the real-time data and the static standing average value obtained previously to obtain the difference on each axis. These differences reflect the posture changes that the user has experienced from the standing state to the present. In this way, the degree of deviation between the user's posture and the baseline standing posture at each time point can be quantified, and the degree of deviation can also be called posture deviation. To ensure the accuracy and stability of the posture deviation, the system smoothes the posture deviation over a period of time, considering that the accelerometer may be affected by environmental noise. This processing method is similar to averaging, but more complicated: instead of simply averaging all the data, it selects the posture deviations of the last 10 time points and calculates their average. This helps reduce the impact of instantaneous jitter or short-term anomalies on the results. After that, the smoothed posture deviation will be stored in a special queue, which only stores the latest data within 3 seconds. Whenever a new posture deviation is calculated, the oldest data in the queue will be replaced to ensure that the information in the queue is always up to date. When the system has accumulated enough time series posture deviations, it can start to analyze whether there are abnormal patterns in these data. For example, if the posture deviation is found to increase significantly over a period of time and this state persists, it may mean that the user is performing abnormal actions, such as falling suddenly. However, some normal activities in daily life, such as squatting and standing up, can also cause a temporary increase in posture deviation. Therefore, the system not only pays attention to the posture deviation itself, but also considers its trend over time. If the increase in posture deviation is accompanied by lower acceleration fluctuations, that is, the user's posture becomes relatively stable, the system will further use the indoor positioning module to confirm the user's position to determine whether a fall has actually occurred. In this way, the system can distinguish between common posture changes in daily activities and real fall events, and decide whether to trigger an alarm.

[0030] Then, considering that the time queue of posture deviation is time series data, there is a temporal order and mutual dependence relationship between its various data points. Therefore, in order to capture and dig out the correlation influence between the data at various time points, in the technical solution of the present application, a sequence encoder based on forward LSTM is used to perform time series encoding on the time queue of the posture deviation to better capture the behavior pattern within a long time span, and obtain the implicit coding vector of the posture deviation time series feature. In other words, forward LSTM is particularly suitable for processing such time series data with long-term dependencies. By introducing structures such as cell states, input gates, forget gates, and output gates, it can better control the flow of information in the time series and remember long-term information, thereby accurately modeling the complex patterns and dependencies in the posture deviation time series, so as to better characterize the dynamic patterns and changing trends of the posture deviation features. Accordingly, the S4 includes: using a sequence encoder based on forward LSTM to perform time series encoding on the time queue of the posture deviation to obtain the implicit coding vector of the posture deviation time series feature.

[0031] Subsequently, each user's real-time posture data contains information of multiple dimensions (such as components of acceleration in different directions, etc.), and these components usually have higher dimensions, and there is a certain correlation between them. Based on this, in the technical solution of the present application, a posture embedding encoder based on a fully connected layer is used to fully connect the real-time posture data of each user in the time queue of the real-time posture data of the user to map these high-dimensional data to a lower-dimensional space, and obtain the time queue of the real-time posture embedding coding vector of the user. That is, through the fully connected coding, the potential relationship between different dimensions can be excavated to capture richer and more detailed features in the posture data. Correspondingly, the S5 includes: using a posture embedding encoder based on a fully connected layer to fully connect the real-time posture data of each user in the time queue of the real-time posture data of the user to obtain the time queue of the real-time posture embedding coding vector of the user.

[0032] Furthermore, considering that each user's real-time posture embedding coding vector directly has an associated influence relationship on a time scale, such as when walking normally and falling, the rhythm and amplitude of posture changes are different. Therefore, in order to capture the relationship between the posture embedding vectors of adjacent and non-adjacent time points, discover the potential pattern of posture changes, and thus better understand the dynamic changes of the entire posture, the present application performs temporal context encoding of the posture jump degree on the time queue of the user's real-time posture embedding coding vector to obtain an implicit coding vector of the posture change temporal pattern. In particular, the temporal context encoding of the posture jump degree can transmit information across time steps, capture long-term and short-term dependencies, and comprehensively analyze the posture change process to avoid ignoring important information due to only considering short-term dependencies.

[0033] Then, if Figure 2 As shown, in step S6, the S6 includes: S61, calculating the message propagation significance factor of each user real-time posture embedded coding vector in the time queue of the user real-time posture embedded coding vector to obtain the time queue of the user real-time posture message propagation significance factor; S62, performing gated propagation on the time queue of the user real-time posture message propagation significance factor to obtain the time queue of the user real-time posture message propagation significance weight; S63, using the time queue of the user real-time posture message propagation significance weight, calculating the weighted sum of the time queue of the user real-time posture embedded coding vector to obtain the implicit coding vector of the posture change timing pattern.

[0034] Among them, Figure 3 As shown, in step S61, the message propagation significance factor of each user real-time posture embedded coding vector in the time queue of the user real-time posture embedded coding vector is calculated to obtain the time queue of the user real-time posture message propagation significance factor, including: S611, calculating the feature jump degree of each user real-time posture embedded coding vector in the time queue of the user real-time posture embedded coding vector to obtain the time queue of the user real-time posture feature jump degree; S612, calculating the message propagation space span of each user real-time posture embedded coding vector in the time queue of the user real-time posture embedded coding vector to obtain the time queue of the user real-time posture message propagation space span; S613, based on the feature jump degree and message propagation space span of each user real-time posture embedded coding vector in the time queue of the user real-time posture embedded coding vector, calculating the message propagation significance factor of each user real-time posture embedded coding vector to obtain the time queue of the user real-time posture message propagation significance factor.

[0035] Among them, Figure 4 As shown, in step S62, the time queue of the user's real-time posture message propagation significance factor is gated and propagated to obtain the time queue of the user's real-time posture message propagation significance weight, including: S621, the time queue of the user's real-time posture message propagation significance factor is normalized based on the softmax function to obtain the time queue of the normalized user real-time posture message propagation significance factor; S622, the time queue of the normalized user real-time posture message propagation significance factor is screened based on the mask function to obtain the time queue of the user's real-time posture message propagation significance weight.

[0036] Specifically, the degree of feature change between nodes is first quantified. This step is achieved by calculating the jump degree of the user real-time posture feature of each user real-time posture embedding encoding vector in the time queue of the user real-time posture embedding encoding vector. The jump degree of the user real-time posture feature reflects the distance or difference in the node feature space, which helps to capture local structural information and mutation characteristics between nodes. The jump degree of the user real-time posture feature can reveal the important characteristics of each node feature in the feature sequence distribution, such as boundaries, outliers, etc., so as to not only identify which node features have large changes in the network, but also emphasize the nodes that may have an important impact on the propagation dynamics.

[0037] The process can be expressed as follows: ; ; ; in, is the time queue of the user's real-time posture embedding encoding vector, , , and are the first, second, and third time series of the user's real-time posture embedding encoding vector. and The user's real-time posture embedding coding vector, yes Middle The eigenvalues ​​at the positions, yes The number of eigenvalues ​​in , and They are the time queues of the weighted average feature values ​​of the user's real-time posture embedding features. and The weighted average feature value of the user's real-time posture embedding feature, It is the first in the time queue of the user's real-time posture feature jump degree. The real-time gesture feature jump degree of each user.

[0038] Next, when evaluating the ability of information to spread in the network, the system calculates the user real-time posture message propagation space span of each user real-time posture embedding encoding vector in the time queue of the user real-time posture embedding encoding vector to obtain the user real-time posture message propagation space span. This step goes beyond simple adjacency and takes into account broader connectivity and path length. The user real-time posture message propagation space span helps identify long-distance dependencies and critical paths for information flow.

[0039] The process can be expressed as follows: ; in, and They represent the time queue of the user's real-time posture embedding encoding vector. and The user's real-time posture embedding encoding vector, express arrive The number of times the message is spread, It is the first in the time queue of the user's real-time posture message propagation space span. The spatial span of real-time posture message propagation of each user.

[0040] Combining the information obtained in the above two steps, the next step is to calculate the user real-time posture message propagation saliency factor of each user real-time posture embedding encoding vector to form a time queue of the user real-time posture message propagation saliency factor. This stage comprehensively considers the user real-time posture feature jump degree and the user real-time posture message propagation spatial span to determine which nodes play a more important role in the information propagation process. The user real-time posture message propagation saliency factor is similar to the attention mechanism, in which nodes adjust the way information is exchanged according to the importance of themselves and their neighbors. In this way, it ensures that both important nodes can be highlighted and the overall structural information is retained, and each node is assigned a score that reflects its importance in the information propagation process.

[0041] The process can be expressed as follows: ; in, It is the first in the time queue of the user's real-time posture feature jump degree. The real-time gesture feature jump degree of each user, It is the first in the time queue of the user's real-time posture message propagation space span. The spatial span of the real-time posture message propagation of each user, and is the modulation parameter, It is the first in the time queue of the significant factor of user real-time posture message propagation The significant factor of user real-time posture message propagation.

[0042] Subsequently, the time queue of the user's real-time posture message propagation significance factor is input into the gated transfer unit to obtain the time queue of the user's real-time posture message propagation significance weight. The gated transfer unit plays the role of regulating the information flow here, and controls which information should be emphasized or suppressed through the learned user's real-time posture message propagation significance weight. The gating mechanism borrows the concepts of forget gate and input gate in recurrent neural network, allowing the model to selectively remember or ignore part of the input. Accordingly, the gated transfer unit can enhance the attention to different node features by introducing additional parameters, thereby improving the expressiveness of the model. Such a design can not only adjust the information flow, but also ensure that only the most relevant information is passed, while less relevant features are appropriately weakened. Finally, based on the user's real-time posture message propagation significance weight obtained in the previous step, the weighted sum of the time queue of the user's real-time posture embedded coding vector is calculated to obtain the implicit coding vector of the posture change timing pattern.

[0043] The process can be expressed as follows: ; ; in, It is the first in the time queue of the significant factor of user real-time posture message propagation The significant factor of user real-time posture message propagation, is the normalization function, is a masking operation, is a predetermined threshold, It is the first in the time queue of the significant weight of the user's real-time posture message propagation The significant weight of the real-time posture message propagation of each user, represents the time queue of the user's real-time posture embedding encoding vector The user's real-time posture embedding encoding vector, is the number of vectors in the time queue of the user's real-time posture embedding encoding vector, is the implicit encoding vector of the posture change timing pattern.

[0044] Then, considering that the implicit coding vector of the posture change timing pattern mainly reflects the dynamic change pattern of the posture in the time series, it reflects the continuity and evolution trend of the action. The implicit coding vector of the posture offset timing feature focuses on reflecting the change of the deviation degree of the real-time posture relative to the static standing posture over time. These two vectors describe the user posture information from different angles. If only one feature is relied on, it may not be possible to accurately judge the fall situation. For example, the posture change pattern may be similar in some violent exercises and falls, but the posture deviation may be significantly different. Based on this, in the technical solution of the present application, the implicit coding vector of the posture change timing pattern and the implicit coding vector of the posture offset timing feature are spliced ​​to obtain the user posture timing multi-dimensional implicit coding vector. In this way, the obtained user posture timing multi-dimensional implicit coding vector combines the time pattern of posture change and the characteristics of posture offset to form a more comprehensive and detailed feature representation. This representation not only includes the dynamic characteristics of user behavior, but also reflects its changes relative to the baseline posture, providing a solid foundation for subsequent analysis.

[0045] Afterwards, the user posture temporal multi-dimensional implicit coding vector is input into a state recognizer based on a classifier to obtain a recognition result. That is, the user posture temporal multi-dimensional implicit coding vector obtained by splicing the posture change temporal pattern implicit coding vector and the posture offset temporal feature implicit coding vector is used for classification processing, so that the classifier learns the feature combination corresponding to different posture patterns according to the features represented by each dimension in the vector, thereby accurately classifying them into different state categories to obtain a recognition result.

[0046] Correspondingly, the S8 includes: inputting the user posture temporal multi-dimensional implicit coding vector into a classifier-based state identifier to obtain the recognition result; in response to the recognition result that the user falls, generating a fall warning prompt.

[0047] Preferably, inputting the user gesture temporal multi-dimensional implicit coding vector into a classifier-based state recognizer to obtain the recognition result comprises: Determine the multi-dimensional implicit coding vector of the user posture time series input based on the state identifier of the classifier to obtain the multi-dimensional implicit probability value of the user posture time series , the multi-dimensional implicit probability value of the user posture time series represents the probability of the user falling; The characteristic mean of the multi-dimensional implicit coding vector of the user posture time series is multiplied by the multi-dimensional implicit probability value of the user posture time series to obtain the multi-dimensional implicit statistical field value of the user posture time series. ;in, represents the feature mean of the multi-dimensional implicit encoding vector of the user posture temporal sequence, represents the multi-dimensional implicit probability value of the user posture time series, Representing the multi-dimensional implicit statistical field value of the user posture time series; Subtract one from the multi-dimensional implicit statistical field value of the user posture time series and divide it by the multi-dimensional implicit statistical field value of the user posture time series to obtain the multi-dimensional implicit partial probability value of the user posture time series ;in, represents the multi-dimensional implicit statistical field value of the user posture time series, Indicates the multi-dimensional implicit partial probability value of the user posture time series; Calculate the multi-dimensional implicit coding vector of the user posture time series with the multi-dimensional implicit partial probability value of the user posture time series as the power function of the exponent , and multiply it by the multi-dimensional implicit partial probability value of the user posture time series to obtain the multi-dimensional implicit microscopic representation vector of the user posture time series ;in, represents the temporal multi-dimensional implicit coding vector of the user posture, represents the multi-dimensional implicit partial probability value of the user posture time series, represents a power function in which the multi-dimensional implicit coding vector of the user posture temporal sequence is exponentially expressed by the multi-dimensional implicit partial probability value of the user posture temporal sequence, represents dot product, A multi-dimensional implicit microscopic representation vector representing the temporal sequence of the user's posture; After multiplying the multi-dimensional implicit coding vector of the user posture time series by the multi-dimensional implicit partial probability value of the user posture time series, an exponential function with a natural constant as the base is calculated to obtain the multi-dimensional implicit macroscopic mapping vector of the user posture time series ;in, represents the temporal multi-dimensional implicit coding vector of the user posture, represents the multi-dimensional implicit partial probability value of the user posture time series, represents dot product, It means to calculate the exponential function with natural constant as base. A multi-dimensional implicit macroscopic mapping vector representing the temporal sequence of the user's posture; After calculating the base 2 logarithm of the multi-dimensional implicit microscopic representation vector of the user posture temporal sequence, the logarithm of the multi-dimensional implicit macroscopic representation vector of the user posture temporal sequence is weighted and summed to obtain the optimized multi-dimensional implicit encoding vector of the user posture temporal sequence. ;in, represents the multi-dimensional implicit microscopic representation vector of the user's posture temporal sequence, represents the multi-dimensional implicit macroscopic mapping vector of the user's posture temporal sequence, represents dot product, Indicates point addition, and is the weighted modulation parameter, A multi-dimensional implicit encoding vector representing the optimized user posture temporal sequence; The optimized user posture temporal multi-dimensional implicit coding vector is input into a classifier-based state recognizer to obtain a recognition result.

[0048] Since each user real-time posture embedded coding vector in the time queue of the user real-time posture embedded coding vector represents the embedded coding features of the user real-time posture data in the local time domain, when performing feature sequence message transmission based on feature jump degree, the embedded coding features in the local time domain will have a micro jump degree-macro sequence message transmission deviation relative to the global time domain, and when the posture change timing pattern implicit coding vector and the posture offset timing feature implicit coding vector are spliced, a local-global feature space deviation relative to the posture deviation time domain feature space is further introduced, so that when the user posture timing multi-dimensional implicit coding vector is input into the state recognizer based on the classifier, a dynamic deviation of the feature to class target probability mapping is caused, thereby reducing the accuracy of the obtained recognition result.

[0049] Therefore, the partial low-order derivatives of the statistical distribution field corresponding to the user posture temporal multidimensional implicit coding vector are used as non-overlapping macro-feature representation behavior patches of the user posture temporal multidimensional implicit coding vector, and the different macro-behavior patch organization spaces under the non-isotropic backbone structure of the user posture temporal multidimensional implicit coding vector are used to enhance the dynamic sensitivity of the long-series micro-complex information distribution of the user posture temporal multidimensional implicit coding vector to the class probability macro-representation behavior, thereby promoting the iterative dynamic consistency of the class target between the classification target and the extracted features during the feature space-class probability mapping, so as to improve the accuracy of the recognition result obtained by the state recognizer based on the classifier input of the user posture temporal multidimensional implicit coding vector.

[0050] Then, in response to the recognition result that the user has fallen, a fall warning prompt is generated. It should be understood that a person's fall may cause serious physical injury, especially for specific groups such as the elderly and patients. Timely detection and response to fall events can minimize the risk of injury. The purpose of generating a fall warning prompt is to take prompt measures to protect the user's life, health and safety when a fall is detected.

[0051] It is worth mentioning that after confirming the fall, the system takes action immediately. The first is a local alarm: the server sends a signal to the accelerometer worn by the elderly, activating the built-in buzzer to sound an alarm to alert people nearby. At the same time, the server also notifies the wireless image sensor module installed at home. The wireless image sensor module has a camera that has the function of listening to the sound and can locate the elderly according to the sound of the buzzer and take a photo. This photo is then transmitted to the local server as an important basis for subsequent assessment of the situation. In order to provide further help information, the environmental perception module will also be activated to collect environmental parameters such as indoor temperature and humidity, carbon monoxide concentration, smoke concentration, and PM2.5 concentration. These data are very important for doctors and can help them make a preliminary judgment on the cause of the fall, such as whether there is a gas leak or air quality problem. All the collected information - including the time of the fall, the photos taken, and the environmental conditions - will be integrated into a detailed report by the local server. Finally, this report containing key information is pushed to the preset remote client, such as the mobile phone or computer of family members or medical staff, via the Internet. In this way, even those who are not at the scene can understand the elderly person’s situation at the first time and can respond promptly, such as contacting emergency services or going to check on them. Throughout the entire process, from detecting falls to generating early warning prompts and notifying relevant personnel, the system responds very quickly, ensuring that necessary rescue measures can be taken quickly in an emergency, thereby protecting the health and safety of users.

[0052] Based on the above embodiments, see Figure 5The figure is a schematic diagram of the structure of a person fall alarm device 100 based on a wireless sensor network in an embodiment of the present application. The person fall alarm device 100 based on a wireless sensor network includes: a data acquisition module 110, which is used to prompt the user to stand steadily and obtain static standing posture data when starting; a collection and calculation module 120, which is used to collect the user's real-time posture data to obtain a time queue of the user's real-time posture data, and calculate the time queue of posture deviation based on the static standing posture data; a time series encoding module 130, which is used to perform time series encoding on the time queue of posture deviation to obtain an implicit encoding vector of posture offset time series feature; an embedded encoding module 140, which is used to perform time series encoding on each user's real-time posture data in the time queue of the user's real-time posture data Embedding coding to obtain a time queue of the user's real-time posture embedded coding vector; a temporal context coding module 150, used to perform temporal context coding of the posture jump degree on the time queue of the user's real-time posture embedded coding vector to obtain an implicit coding vector of the posture change timing pattern; a vector splicing module 160, used to splice the implicit coding vector of the posture change timing pattern and the implicit coding vector of the posture offset timing feature to obtain a user posture timing multi-dimensional implicit coding vector; a fall warning prompt module 170, used to obtain a recognition result based on the user posture timing multi-dimensional implicit coding vector, and generate a fall warning prompt based on the recognition result. Among them, the acquisition and calculation module 120 is used to: calculate the posture deviation of the user's real-time posture data at each predetermined time point in the time queue of the user's real-time posture data relative to the static standing posture data to obtain the time queue of the posture deviation.

[0053] Here, those skilled in the art can understand that the specific functions and operations of each module in the above-mentioned personnel fall alarm device 100 based on the wireless sensor network have been referred to above. Figures 1 to 4 The invention has been introduced in detail in the description of the personnel fall alarm method based on wireless sensor network, and therefore, its repeated description will be omitted.

[0054] Those skilled in the art will appreciate that all or part of the steps in the above method can be completed by instructing related hardware through a program, and the program can be stored in a computer-readable storage medium, such as a read-only memory, a disk or an optical disk. Optionally, all or part of the steps in the above embodiment can also be implemented using one or more integrated circuits. Accordingly, each module / unit in the above embodiment can be implemented in the form of hardware or in the form of software function modules. The present application is not limited to any particular form of combination of hardware and software.

[0055] Unless otherwise defined, all terms (including technical and scientific terms) used herein have the same meaning as commonly understood by those of ordinary skill in the art to which this application belongs. It should also be understood that terms such as those defined in common dictionaries should be interpreted as having a meaning consistent with their meaning in the context of the relevant technology, and should not be interpreted in an idealized or extremely formal sense, unless explicitly defined as such herein.

[0056] The above is an explanation of the present application and should not be considered as limiting thereof. Although several exemplary embodiments of the present application have been described, those skilled in the art will readily appreciate that many modifications may be made to the exemplary embodiments without departing from the novel teachings and advantages of the present application.

Claims

1. A personnel fall alarm method based on wireless sensor network, characterized in that: include: S1: Wear the acceleration sensor module and place the indoor positioning module; S2: When starting, prompt the user to stand steadily and obtain static standing posture data; S3: collecting the user's real-time posture data to obtain a time queue of the user's real-time posture data, and calculating a time queue of posture deviation based on the static standing posture data; S4: performing time series coding on the time queue of the posture deviation to obtain an implicit coding vector of the posture deviation time series feature; S5: performing embedding coding on each user real-time posture data in the time queue of the user real-time posture data to obtain a time queue of user real-time posture embedding coding vectors; S6: performing temporal context coding of the gesture jump degree on the time queue of the user's real-time gesture embedded coding vector to obtain a gesture change temporal pattern implicit coding vector; S7: concatenating the posture change timing pattern implicit coding vector and the posture offset timing feature implicit coding vector to obtain a user posture timing multi-dimensional implicit coding vector; S8: Obtain a recognition result based on the multi-dimensional implicit coding vector of the user posture time series, and generate a fall warning prompt based on the recognition result.

2. A method for alarming a person falling down based on a wireless sensor network according to claim 1, characterized in that: Based on the static standing posture data, a time queue of posture deviation is calculated, including: calculating the posture deviation of the user's real-time posture data at each predetermined time point in the time queue of the user's real-time posture data relative to the static standing posture data to obtain the time queue of the posture deviation.

3. A method for alarming a person falling down based on a wireless sensor network according to claim 2, characterized in that: The S4 includes: using a forward LSTM-based sequence encoder to perform time series encoding on the time queue of the posture deviation to obtain the implicit encoding vector of the posture deviation time series feature.

4. A method for alarming a person falling down based on a wireless sensor network according to claim 3, characterized in that: The S5 includes: using a posture embedding encoder based on a fully connected layer to perform fully connected encoding on each user real-time posture data in the time queue of the user real-time posture data to obtain the time queue of the user real-time posture embedding coding vector.

5. A method for alarming a person falling down based on a wireless sensor network according to claim 4, characterized in that: The S6 comprises: Calculate the message propagation significance factor of each user real-time posture embedding coding vector in the time queue of the user real-time posture embedding coding vector to obtain the time queue of the user real-time posture message propagation significance factor; Perform gated propagation on the time queue of the user's real-time posture message propagation significance factor to obtain the time queue of the user's real-time posture message propagation significance weight; The weighted sum of the time queues of the user's real-time posture embedded coding vectors is calculated based on the time queues of the user's real-time posture message propagation significance weights to obtain the implicit coding vector of the posture change timing pattern.

6. A method for alarming a person falling down based on a wireless sensor network according to claim 5, characterized in that: Calculating the message propagation significance factor of each user real-time posture embedding coding vector in the time queue of the user real-time posture embedding coding vector to obtain the time queue of the user real-time posture message propagation significance factor, including: Calculate the feature jump degree of each user real-time posture embedding coding vector in the time queue of the user real-time posture embedding coding vector to obtain the time queue of the user real-time posture feature jump degree; Calculating the message propagation space span of each user real-time posture embedding coding vector in the time queue of the user real-time posture embedding coding vector to obtain the time queue of the user real-time posture message propagation space span; Based on the characteristic jump degree and message propagation space span of each user real-time posture embedding coding vector in the time queue of the user real-time posture embedding coding vector, the message propagation significance factor of each user real-time posture embedding coding vector is calculated to obtain the time queue of the user real-time posture message propagation significance factor.

7. A method for alarming a person falling down based on a wireless sensor network according to claim 6, characterized in that: Performing gated propagation on the time queue of the user's real-time posture message propagation significance factor to obtain the time queue of the user's real-time posture message propagation significance weight, including: Performing a normalization process based on a softmax function on the time queue of the user's real-time posture message propagation significance factor to obtain a normalized time queue of the user's real-time posture message propagation significance factor; The time queue of the normalized user real-time posture message propagation significance factor is screened based on a mask function to obtain the time queue of the user real-time posture message propagation significance weight.

8. A method for alarming a person falling down based on a wireless sensor network according to claim 7, characterized in that: The S8 comprises: Inputting the user posture temporal multi-dimensional implicit coding vector into a classifier-based state recognizer to obtain the recognition result; In response to the recognition result that the user has fallen, a fall warning prompt is generated.

9. A personnel fall alarm device based on a wireless sensor network, characterized in that: include: A data acquisition module, used to prompt the user to stand steadily and obtain static standing posture data when starting; A collection and calculation module, used for collecting the user's real-time posture data to obtain a time queue of the user's real-time posture data, and calculating a time queue of posture deviation based on the static standing posture data; A time series encoding module, used for performing time series encoding on the time queue of the posture deviation to obtain an implicit encoding vector of the posture deviation time series feature; An embedding coding module, used for embedding coding each user real-time posture data in the time queue of the user real-time posture data to obtain a time queue of user real-time posture embedding coding vectors; A temporal context coding module, used for performing temporal context coding of the gesture jump degree on the time queue of the user's real-time gesture embedding coding vector to obtain an implicit coding vector of the gesture change temporal pattern; A vector concatenation module, used for concatenating the posture change timing pattern implicit coding vector and the posture offset timing feature implicit coding vector to obtain a user posture timing multi-dimensional implicit coding vector; The fall warning prompt module is used to obtain a recognition result based on the multi-dimensional implicit coding vector of the user posture time sequence, and generate a fall warning prompt based on the recognition result.

10. The personnel fall alarm device based on wireless sensor network according to claim 9, characterized in that: The collection and calculation module is used to calculate the posture deviation of the user's real-time posture data at each predetermined time point in the time queue of the user's real-time posture data relative to the static standing posture data to obtain the time queue of the posture deviation.

Images