Intelligent home voice interaction control system and method based on nursing bed

Abstract

The invention provides a smart home voice interaction control system and method based on a nursing bed, and the system comprises a data collection module which is used for collecting user sign data, environment data and bed body posture data in real time; the voice receiving module is used for collecting voice data in real time and carrying out noise reduction processing on the voice data to obtain standard voice data; the multi-modal fusion module is used for performing multi-modal fusion according to the time dimension based on the user physical sign data, the environment data, the bed body posture data and the standard voice data to generate a target scene vector; and the control module is used for determining target home equipment and a control instruction for the equipment based on the target scene vector, and controlling the target home equipment according to the control instruction. Therefore, great convenience is provided for the user, effective integration of the nursing bed and the intelligent household equipment is also improved, and convenient interaction between the nursing bed and the intelligent household equipment is also improved.

Description

Technical Field

[0001] This invention relates to the field of equipment control technology, and in particular to a smart home voice interaction control system and method based on a nursing bed. Background Technology

[0002] Currently, with the increasing aging of society and the growing number of people with special care needs, nursing beds are being used more and more widely in homes and medical institutions. However, traditional nursing beds have relatively limited functions, mostly only meeting basic lying needs. They lack effective integration with smart home devices and convenient interaction methods. For example, when patients or the elderly use nursing beds, they may need to adjust the brightness and temperature of indoor lights, call for caregivers, etc. Currently, they often need to use multiple different devices or operate manually, which is very inconvenient for users with limited mobility. Therefore, in order to overcome the above-mentioned defects, the present invention provides a smart home voice interaction control system and method based on a nursing bed. Summary of the Invention

[0003] This invention provides a smart home voice interaction control system and method based on a nursing bed. It collects user vital signs data, environmental data, bed posture data, and voice data, and performs noise reduction processing on the collected voice data to effectively determine accurate voice signals. Next, it performs multimodal fusion of the obtained user vital signs data, environmental data, bed posture data, and standard voice data according to the time dimension to accurately and effectively determine the target scenario vector, thereby determining the user's comprehensive state at different times. Finally, based on the determined target scenario vector, it determines the target home devices and the control commands for the devices, ultimately enabling control of the target home devices according to the control commands. This provides great convenience for users, improves the effective integration of the nursing bed and smart home devices, and enhances the convenient interaction between the nursing bed and smart home devices.

[0004] This invention provides a smart home voice interaction control system based on a nursing bed, comprising: The data acquisition module is used to collect user vital signs data, environmental data, and bed posture data in real time. The voice receiving module is used to collect voice data in real time and perform noise reduction processing on the voice data to obtain standard voice data. The multimodal fusion module is used to generate a target scenario vector by performing multimodal fusion based on user vital signs data, environmental data, bed posture data, and standard voice data according to the time dimension. The control module is used to determine the target home appliances and control commands for the appliances based on the target scenario vector, and to control the target home appliances according to the control commands.

[0005] Preferably, a smart home voice interaction control system based on a nursing bed includes a data acquisition module comprising: The vital signs data acquisition unit is used to collect user vital signs data in real time based on the vital signs sensor array, and to perform initial storage of the user vital signs data; An environmental data acquisition unit is used to collect environmental data in real time from the area where the nursing bed is located based on environmental sensors, and to store the environmental data in a second location. The bed posture data acquisition unit is used to acquire linear displacement pulse signals in real time based on the sensing devices inside the nursing bed actuator, and to convert the acquired linear displacement pulse signals into posture data according to the data processing center integrated into the nursing bed to obtain bed posture data.

[0006] Preferably, a smart home voice interaction control system based on a nursing bed includes a voice receiving and analysis module, comprising: The voice acquisition unit is used to acquire external voice data sequences in real time. The keyword matching unit is used to obtain preset voice keywords and match the preset voice keywords with external voice data sequences. When the preset voice keywords are present in the external voice data sequence in the matching result, a voice receiving command is generated. The voice acquisition unit is also used to take the time point corresponding to the preset voice keywords as the starting timestamp, and to acquire key voice data in real time according to the starting timestamp and the voice receiving instruction. The filtering unit is used to filter key speech data to obtain standard speech data.

[0007] Preferably, a smart home voice interaction control system based on a nursing bed includes a filtering unit, comprising: The audio acquisition subunit is used to acquire ambient noise audio from different sound sources in the nursing environment. The learning sub-unit is used to learn the environmental noise audio from different sound sources, determine the environmental noise audio characteristics of each sound source, and construct a noise recognition network based on the environmental noise audio characteristics of each sound source. The noise management strategy allocation subunit is used to acquire noise attributes, including steady-state noise and sudden speech interference noise, and to allocate different noise management strategies according to the noise attributes. The noise management mechanism determination subunit is used to associate the noise identification network with the noise management strategy to generate the noise management mechanism; The first processing subunit is used to input the key voice data collected in real time into the noise management mechanism for analysis, execute the corresponding noise management strategy according to the analysis results, and obtain the first audio information according to the execution results. The coherence determination subunit is used for: The first audio information is amplified and processed to obtain the second audio information, and the information content of the second audio information is identified and the coherence of the information content is judged. When the coherence of the information content reaches the preset standard, the second audio information will be output as the standard speech information. The second processing subunit is used to analyze the second audio information based on a preset semantic model when the coherence of the information content does not meet the preset standard, and to supplement the information content of the second audio information according to the analysis results until the supplemented information content meets the preset standard, and then use the supplemented second audio information as the standard speech information.

[0008] Preferably, a smart home voice interaction control system based on a nursing bed includes a first processing subunit comprising: The identification subunit is used to input key speech data into the noise management mechanism for identification based on the noise identification network, and to determine the noise data and noise source of the key speech data. The strategy allocation subunit is used for: Obtain the noise attributes of noise sources and assign noise management strategies to denoise key speech data based on the noise attributes; When the noise source is a steady-state noise, the corresponding noise management strategy is to filter the noise data corresponding to the noise source based on a preset filtering model. When the noise source is characterized as sudden speech interference noise, the corresponding noise management strategy is to instantly reduce the gain of the air microphone to reduce the noise decibels corresponding to the noise data.

[0009] Preferably, a smart home voice interaction control system based on a nursing bed includes: the smart nursing bed and home devices are connected based on a communication protocol, and the system is connected to the Internet based on a preset gateway.

[0010] Preferably, a smart home voice interaction control system based on a nursing bed includes a multimodal fusion module, comprising: The data retrieval unit is used to retrieve the collected user vital signs data, environmental data, bed posture data, and standard voice data, and to extract the timestamps corresponding to the user vital signs data, environmental data, bed posture data, and standard voice data. Multimodal fusion unit, used for: Based on timestamps, user vital signs data, environmental data, bed posture data, and standard voice data under the same time dimension are time-series aligned, and a multimodal data stream with a unified timestamp is constructed based on the time-series alignment results. Feature extraction is performed on the multimodal data stream, and feature vectors for each data modality are obtained based on the feature extraction. The interaction relationships between multimodal data streams are determined based on a preset control protocol, and the feature vectors of each data modality are weighted and fused based on the interaction relationships and a preset attention mechanism. The target scenario vector is obtained based on the weighted fusion result.

[0011] Preferably, a smart home voice interaction control system based on a nursing bed includes a control module comprising: Equipment determination unit, used for: The target scenario vector is obtained, and its dimensions are analyzed to determine the multi-dimensional feature vectors contained in the target scenario. The pre-trained vector device mapping model is used to analyze and identify multi-dimensional feature vectors, and the target home device is determined based on the analysis and identification results. Instruction generation unit, used for: The system obtains control command elements corresponding to the target home devices from the management terminal. At the same time, it determines the motion control parameters for each target home device based on the target scenario vector, and performs parameter adaptation on the execution command elements in the control command elements based on the motion control parameters. Control commands for the target home appliances are obtained based on the parameter adaptation results; The device control unit is used to control target home appliances based on control commands.

[0012] Preferably, a smart home voice interaction control system based on a nursing bed includes a control module comprising: The data collection unit is used to acquire historical voice interaction control data between the user and home appliances within a target time period, as well as the scene information corresponding to each historical voice interaction control, and to extract the scene features corresponding to the scene information. Service strategy library building unit, used for: Based on scene features, the scene information corresponding to the historical voice interaction control data is classified into scenes, and scene recognition labels are generated for each category based on scene features. The scene classification results are then marked based on the scene recognition labels. Based on the labeling results, the historical voice interaction control data for each category of scenario is traversed, and the control time, control items and corresponding control quantification indicators for each historical voice interaction control data are determined based on the data traversal results. Construct a data control group for the same control project under the same category of scenarios, and determine the fluctuation range of the control quantitative indicators based on the data control group; Based on the fluctuation range, the action amplitude of each control item is limited. At the same time, based on the proportion of different control quantitative indicators in the fluctuation range, the user's control preference is determined. Based on the control time, action amplitude, and control preference, the user's control habits for each control item in different scenarios are obtained. A service strategy library is built based on users' control habits for various control items in different scenarios; Automatic service unit, used for: The system monitors the user's environment in real time and proactively sends service requests to the user when the time and environment meet the service policies already in the service policy library. Based on proactively sending results to monitor user feedback instructions, and upon receiving user feedback instructions, retrieve the target service policy corresponding to the current time node and scenario from the service policy library; Active control of home appliances based on target service strategies.

[0013] This invention provides a smart home voice interaction control method based on a nursing bed, comprising: Step 1: Collect user vital signs data, environmental data, and bed posture data in real time; Step 2: Collect voice data in real time and perform noise reduction processing on the voice data to obtain standard voice data; Step 3: Based on user vital signs data, environmental data, bed posture data, and standard speech data, multimodal fusion is performed according to the time dimension to generate a target scenario vector; Step 4: Determine the target home appliances and control commands for the appliances based on the target scenario vector, and control the target home appliances according to the control commands.

[0014] Compared with the prior art, the beneficial effects of the present invention are as follows: By collecting user vital signs data, environmental data, bed posture data, and voice data, and performing noise reduction processing on the collected voice data, accurate voice signals can be effectively determined. Secondly, the obtained user vital signs data, environmental data, bed posture data, and standard voice data are fused using a multimodal approach based on the time dimension to accurately and effectively determine the target scenario vector, thereby determining the user's comprehensive state at different times. Finally, based on the determined target scenario vector, the target home appliances and their control commands are determined, ultimately enabling control of the target home appliances according to the control commands. This provides great convenience for users, improves the effective integration of the nursing bed with smart home devices, and enhances the convenient interaction between the nursing bed and smart home devices.

[0015] Other features and advantages of the invention will be set forth in the description which follows, and will be apparent in part from the description, or may be learned by practicing the invention. The objects and other advantages of the invention may be realized and obtained by means of the structures particularly pointed out in this application.

[0016] The technical solution of the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. Attached Figure Description

[0017] The accompanying drawings are provided to further illustrate the invention and form part of the specification. They are used in conjunction with embodiments of the invention to explain the invention and do not constitute a limitation thereof. In the drawings: Figure 1 This is a structural diagram of a smart home voice interaction control system based on a nursing bed, according to an embodiment of the present invention. Figure 2 This is a structural diagram of the data acquisition module in a smart home voice interaction control system based on a nursing bed, according to an embodiment of the present invention. Figure 3 This is a flowchart of a smart home voice interaction control method based on a nursing bed, as described in an embodiment of the present invention. Detailed Implementation

[0018] The preferred embodiments of the present invention will be described below with reference to the accompanying drawings. It should be understood that the preferred embodiments described herein are for illustration and explanation only and are not intended to limit the present invention.

[0019] Example 1: This example provides a smart home voice interaction control system based on a nursing bed, such as... Figure 1 As shown, it includes: The data acquisition module is used to collect user vital signs data, environmental data, and bed posture data in real time. The voice receiving module is used to collect voice data in real time and perform noise reduction processing on the voice data to obtain standard voice data. The multimodal fusion module is used to generate a target scenario vector by performing multimodal fusion based on user vital signs data, environmental data, bed posture data, and standard voice data according to the time dimension. The control module is used to determine the target home appliances and control commands for the appliances based on the target scenario vector, and to control the target home appliances according to the control commands.

[0020] In this embodiment, the user's vital signs data include data such as the user's heart rate and blood pressure.

[0021] In this embodiment, environmental data refers to humidity data, temperature data, and light data corresponding to the environment.

[0022] In this embodiment, the bed posture data refers to the specific posture of the nursing bed at present, including the tilt angle, etc.

[0023] In this embodiment, noise reduction processing refers to filtering out the noise contained in the speech data, i.e., obtaining standard speech data.

[0024] In this embodiment, multimodal fusion based on the time dimension refers to associating and fusing user vital signs data, environmental data, bed posture data, and standard voice data at the same time. The purpose is to determine the user's overall state at the same moment, thereby facilitating the identification of the target home appliances that need to be controlled.

[0025] In this embodiment, the target scenario vector can accurately represent the overall state of the user at a specific moment, and serve as an input condition to determine the target home device and generate corresponding control commands.

[0026] In this embodiment, the target home appliance refers to the appliance that needs to be controlled.

[0027] The working principle and beneficial effects of the above technical solution are as follows: By collecting user vital signs data, environmental data, bed posture data, and voice data, and performing noise reduction processing on the collected voice data, accurate voice signals can be effectively determined. Secondly, the obtained user vital signs data, environmental data, bed posture data, and standard voice data are fused in a multimodal manner according to the time dimension to accurately and effectively determine the target scenario vector, thereby determining the user's comprehensive state at different times. Finally, based on the determined target scenario vector, the target home appliances and the control commands for the appliances are determined, ultimately enabling the control of the target home appliances according to the control commands. This provides great convenience for users, improves the effective integration of nursing beds and smart home appliances, and enhances the convenient interaction between nursing beds and smart home appliances.

[0028] Example 2: Based on Example 1, this example provides a smart home voice interaction control system based on a nursing bed, such as... Figure 2 As shown, the data acquisition module includes: The vital signs data acquisition unit is used to collect user vital signs data in real time based on the vital signs sensor array, and to perform initial storage of the user vital signs data; An environmental data acquisition unit is used to collect environmental data in real time from the area where the nursing bed is located based on environmental sensors, and to store the environmental data in a second location. The bed posture data acquisition unit is used to acquire linear displacement pulse signals in real time based on the sensing devices inside the nursing bed actuator, and to convert the acquired linear displacement pulse signals into posture data according to the data processing center integrated into the nursing bed to obtain bed posture data.

[0029] In this embodiment, the vital signs sensor array refers to a collection or combination of multiple sensors used to monitor the user's physiological parameters, which may typically include sensors for heart rate, respiration, body temperature, blood pressure, etc.

[0030] In this embodiment, the first storage refers to a storage area or storage unit dedicated to storing user vital sign data.

[0031] In this embodiment, the environmental sensor refers to a sensing device used to monitor the physical environment around the nursing bed, such as a temperature and humidity sensor, a light sensor, and an air quality sensor.

[0032] In this embodiment, the current nursing bed location range refers to the local physical space range where the nursing bed is located, such as the area or enclosed space where the nursing bed is located in a room.

[0033] In this embodiment, the second storage refers to a storage area or storage unit dedicated to storing environmental data.

[0034] In this embodiment, the sensing device inside the nursing bed actuator refers to a sensor installed in the nursing bed's power or transmission device (such as a motor, push rod, etc.) to detect the mechanical motion state.

[0035] In this embodiment, the linear displacement pulse signal refers to the digital pulse signal output by the sensor inside the actuator that corresponds to the mechanical linear displacement. Each pulse usually corresponds to a certain unit of displacement.

[0036] In this embodiment, the data processing center integrated into the nursing bed refers to the embedded hardware or computing module configured inside the nursing bed that is responsible for data calculation and processing.

[0037] In this embodiment, posture data conversion refers to the process of converting linear displacement pulse signals into bed posture data that describes the position or angle of each component of the nursing bed through calculation.

[0038] The beneficial effects of the above technical solution are as follows: by continuously acquiring users' physiological parameters, it provides basic data support for health assessment. At the same time, it dynamically monitors environmental indicators such as temperature and humidity to ensure a comfortable and safe recuperation environment. It also converts high-precision signals and provides real-time feedback on changes in bed morphology. The three work together to form a comprehensive closed-loop monitoring system, which not only ensures the real-time traceability of users' physical condition, but also provides multi-source data fusion support for precise nursing decisions. Ultimately, it realizes proactive early warning and personalized adaptation of intelligent nursing, significantly improving the safety and accuracy of nursing care.

[0039] Example 3: Based on Example 1, this example provides a smart home voice interaction control system based on a nursing bed, including a voice receiving and analysis module, comprising: The voice acquisition unit is used to acquire external voice data sequences in real time. The keyword matching unit is used to obtain preset voice keywords and match the preset voice keywords with external voice data sequences. When the preset voice keywords are present in the external voice data sequence in the matching result, a voice receiving command is generated. The voice acquisition unit is also used to take the time point corresponding to the preset voice keywords as the starting timestamp, and to acquire key voice data in real time according to the starting timestamp and the voice receiving instruction. The filtering unit is used to filter key speech data to obtain standard speech data.

[0040] In this embodiment, the external voice data sequence refers to the original audio signal stream continuously acquired by the voice acquisition unit and arranged in chronological order.

[0041] In this embodiment, preset voice keywords refer to words or phrases that are pre-set by the system to activate the system or represent specific control commands (such as "turn on the light" or "raise the headboard").

[0042] In this embodiment, the voice receiving command refers to a flag signal generated internally by the system when the keyword is successfully matched, which is used to notify subsequent units to start working.

[0043] In this embodiment, the start timestamp refers to the precise time marker at which the system identifies the starting point of a preset voice keyword in the voice data stream.

[0044] In this embodiment, key voice data refers to a continuous voice signal that is captured or recorded starting from the start timestamp, and usually contains complete control command content.

[0045] In this embodiment, standard speech data refers to standardized audio data that meets the input requirements of the speech recognition engine after preprocessing such as noise reduction and echo cancellation of key speech data and filtering.

[0046] The beneficial effects of the above technical solution are as follows: through real-time monitoring and intelligent matching, the start time of effective voice commands can be accurately captured, and key voice segments can be efficiently extracted based on this. Combined with subsequent filtering processing, environmental noise interference can be effectively eliminated, and the clarity and usability of voice data can be greatly improved. It realizes automatic filtering, focusing and enhancement of useful information from massive voice data, and ultimately significantly improves the accuracy of voice recognition and processing and the system response efficiency, providing a reliable data foundation for voice-based interaction and control.

[0047] Example 4: Based on Example 3, this example provides a smart home voice interaction control system based on a nursing bed, including a filtering unit, comprising: The audio acquisition subunit is used to acquire ambient noise audio from different sound sources in the nursing environment. The learning sub-unit is used to learn the environmental noise audio from different sound sources, determine the environmental noise audio characteristics of each sound source, and construct a noise recognition network based on the environmental noise audio characteristics of each sound source. The noise management strategy allocation subunit is used to acquire noise attributes, including steady-state noise and sudden speech interference noise, and to allocate different noise management strategies according to the noise attributes. The noise management mechanism determination subunit is used to associate the noise identification network with the noise management strategy to generate the noise management mechanism; The first processing subunit is used to input the key voice data collected in real time into the noise management mechanism for analysis, execute the corresponding noise management strategy according to the analysis results, and obtain the first audio information according to the execution results. The coherence determination subunit is used for: The first audio information is amplified and processed to obtain the second audio information, and the information content of the second audio information is identified and the coherence of the information content is judged. When the coherence of the information content reaches the preset standard, the second audio information will be output as the standard speech information. The second processing subunit is used to analyze the second audio information based on a preset semantic model when the coherence of the information content does not meet the preset standard, and to supplement the information content of the second audio information according to the analysis results until the supplemented information content meets the preset standard, and then use the supplemented second audio information as the standard speech information.

[0048] In this embodiment, ambient noise audio refers to digital audio signals of background sounds generated by different sound sources in the environment surrounding the nursing bed that may interfere with voice commands.

[0049] In this embodiment, audio features refer to a set of acoustic parameters extracted from the ambient noise audio of a specific sound source that can characterize the unique properties of the noise from that source.

[0050] In this embodiment, the noise recognition network refers to a computational model that is built based on the learned noise audio features and can automatically identify the type of noise in the input audio.

[0051] In this embodiment, noise attributes refer to the classification labels of noise. In this embodiment, noise is divided into two categories based on its time change pattern: steady-state and sudden speech interference.

[0052] In this embodiment, steady-state noise refers to continuous background noise whose intensity and spectral characteristics remain relatively constant over a short period of time.

[0053] In this embodiment, sudden speech interference noise refers to unexpected, sudden, and brief interference sounds that may be similar to speech in the spectrum.

[0054] In this embodiment, the noise management strategy refers to a set of signal processing or suppression schemes pre-designed for identified noise with specific attributes.

[0055] In this embodiment, the noise management mechanism refers to an intelligent noise reduction system that is composed of a noise recognition network and multiple noise management strategies, and can automatically analyze and select strategies.

[0056] In this embodiment, the first audio information refers to the audio of the preliminary noise reduction result after the key voice data has been processed by the noise management mechanism.

[0057] In this embodiment, coherence refers to the level of judgment on the fluency and logical integrity of speech information after semantic analysis.

[0058] In this embodiment, the preset standard refers to the minimum threshold set by the system to determine whether the continuity of voice information is acceptable.

[0059] In this embodiment, the second audio information refers to the audio signal after the first audio information has been amplified by gain.

[0060] In this embodiment, standard speech information refers to the final output speech data that meets the system processing requirements in terms of both clarity and semantic coherence.

[0061] In this embodiment, the preset semantic model refers to a pre-trained contextual language understanding and generation model used to complete incoherent sentences. For example, it can be a sequence-to-sequence model based on Transformer or an autoregressive language model trained on massive amounts of text, represented by the GPT (Generative Pre-trained Transformer) series.

[0062] The beneficial effects of the above technical solution are as follows: by autonomously learning the characteristics of different environmental noises and constructing a recognition network, it can intelligently distinguish between steady-state noise and sudden interference, automatically adapt to the optimal noise reduction strategy, and after targeted processing and signal enhancement of key speech, it can also intelligently determine semantic coherence and automatically complete incomplete information with the help of semantic models, and finally output clear, coherent, high-quality speech, which significantly improves the speech recognition rate and interaction reliability in complex nursing scenarios, and realizes intelligent processing of the entire process from noise reduction to semantic understanding.

[0063] Example 5: Based on Example 4, this example provides a smart home voice interaction control system based on a nursing bed, the first processing subunit including: The identification subunit is used to input key speech data into the noise management mechanism for identification based on the noise identification network, and to determine the noise data and noise source of the key speech data. The strategy allocation subunit is used for: Obtain the noise attributes of noise sources and assign noise management strategies to denoise key speech data based on the noise attributes; When the noise source is a steady-state noise, the corresponding noise management strategy is to filter the noise data corresponding to the noise source based on a preset filtering model. When the noise source is characterized as sudden speech interference noise, the corresponding noise management strategy is to instantly reduce the gain of the air microphone to reduce the noise decibels corresponding to the noise data.

[0064] In this embodiment, noise data refers to the portion of the audio signal that is identified as interfering noise from the key speech data.

[0065] In this embodiment, the noise source refers to the specific object or event type that generates the identified noise data.

[0066] In this embodiment, the preset filtering model refers to a pre-designed or trained digital signal processing algorithm (such as a notch filter for a specific frequency band) specifically designed to filter out steady-state noise.

[0067] In this embodiment, air microphone gain refers to the electronic amplification factor of the system microphone (the hardware that collects audio) on the sound signal in the air.

[0068] In this embodiment, noise decibels refer to the logarithmic unit used to quantify the sound pressure level (i.e., volume) of noise data.

[0069] The beneficial effects of the above technical solution are as follows: by intelligently identifying noise type and sound source, and by adopting filtering and instantaneous gain reduction strategies for steady-state noise and sudden speech interference respectively, a precise and adaptive noise reduction effect is achieved. It can effectively suppress background steady-state noise, while responding quickly to sudden interference, avoiding speech distortion, significantly improving the signal-to-noise ratio and clarity of key speech data, and ultimately ensuring the accuracy of speech information acquisition and the reliability of system interaction in complex environments.

[0070] Example 6: Based on Example 1, this example provides a smart home voice interaction control system based on a nursing bed, including: the smart nursing bed and home devices are connected based on a communication protocol, and the system is connected to the Internet based on a preset gateway.

[0071] The beneficial effects of the above technical solution are as follows: By connecting the nursing bed with home appliances (such as lights, air conditioners, and televisions) through communication protocols, the nursing bed is transformed from an independent device into an intelligent control hub; the user's voice commands or changes in vital signs data can not only control the bed itself, but also automatically link other home appliances to achieve scenario-based responses (for example, if the user says "I'm a little cold," the system can automatically raise the air conditioner temperature and dim the lights); by connecting to the Internet through a preset gateway, the data and control interface of the nursing bed can be extended to the cloud or remote terminals. First, it can utilize more powerful cloud computing resources for data analysis and intelligent decision-making (such as vital sign trend analysis and semantic model updates); second, it allows caregivers or family members to remotely view the user's status, receive alarms, or perform auxiliary control through mobile apps, greatly expanding the spatial and temporal boundaries of nursing care and realizing remote monitoring and intervention.

[0072] Example 7: Based on Example 1, this example provides a smart home voice interaction control system based on a nursing bed, including a multimodal fusion module: The data retrieval unit is used to retrieve the collected user vital signs data, environmental data, bed posture data, and standard voice data, and to extract the timestamps corresponding to the user vital signs data, environmental data, bed posture data, and standard voice data. Multimodal fusion unit, used for: Based on timestamps, user vital signs data, environmental data, bed posture data, and standard voice data under the same time dimension are time-series aligned, and a multimodal data stream with a unified timestamp is constructed based on the time-series alignment results. Feature extraction is performed on the multimodal data stream, and feature vectors for each data modality are obtained based on the feature extraction. The interaction relationships between multimodal data streams are determined based on a preset control protocol, and the feature vectors of each data modality are weighted and fused based on the interaction relationships and a preset attention mechanism. The target scenario vector is obtained based on the weighted fusion result.

[0073] In this embodiment, time alignment refers to arranging and matching data from different sensors or modules (such as vital signs, environment, and voice) with their own time stamps according to a unified time axis to ensure that they are analyzed at the same time or time period.

[0074] In this embodiment, a unified timestamp multimodal data stream refers to integrating various types of data, such as physical characteristics, environment, posture, and voice, after time-series alignment, into a continuous data sequence that shares the same set of timestamps.

[0075] In this embodiment, data modality refers to the type of perception or information to which the data belongs. In this system, it specifically refers to four different data sources or forms: vital signs, environment, bed posture, and voice.

[0076] In this embodiment, the feature vector refers to the numerical representation (usually a number array) extracted from the raw data of a certain modality (such as a heart rate curve or a voice signal) that can characterize its key information.

[0077] In this embodiment, the preset control protocol refers to a set of rules predefined by the system that describes how different modal data affect or are related to each other (for example, changes in bed posture may affect heart rate, and ambient noise levels may affect speech clarity).

[0078] In this embodiment, the interaction relationship refers to the logical association, such as causation, correlation, or triggering, between different modal data features under a specific scenario, as determined by a preset control protocol.

[0079] In this embodiment, the preset attention mechanism refers to a pre-designed or pre-trained algorithm module that can automatically calculate and assign different importance weights to the feature vectors of each modality based on the current context (features of each modality).

[0080] In this embodiment, weighted fusion refers to the process of summing or concatenating the feature vectors of each modality according to the weights allocated by the attention mechanism, thereby synthesizing a unified vector that can represent the overall situation.

[0081] In this embodiment, the target scenario vector refers to a high-dimensional numerical scenario representation that is ultimately obtained through weighted fusion, which integrates information from multiple aspects such as user vital signs, environmental status, bed posture, and voice commands, and is used for subsequent decision-making and control.

[0082] The beneficial effects of the above technical solution are: by integrating multi-source heterogeneous data into a unified data stream through time alignment, and by using feature extraction and intelligent weighted fusion technology, the inherent correlation and complementary information between different modal data are fully explored. The resulting high-precision scenario vector can comprehensively and accurately reflect the user's real-time status and surrounding environment, providing a deeper level of scenario perception and decision support for the intelligent nursing system, and significantly improving the overall system's comprehensive analysis capabilities and response accuracy.

[0083] Example 8: Based on Example 1, this example provides a smart home voice interaction control system based on a nursing bed. The control module includes: Equipment determination unit, used for: The target scenario vector is obtained, and its dimensions are analyzed to determine the multi-dimensional feature vectors contained in the target scenario. The pre-trained vector device mapping model is used to analyze and identify multi-dimensional feature vectors, and the target home device is determined based on the analysis and identification results. Instruction generation unit, used for: The system obtains control command elements corresponding to the target home devices from the management terminal. At the same time, it determines the motion control parameters for each target home device based on the target scenario vector, and performs parameter adaptation on the execution command elements in the control command elements based on the motion control parameters. Control commands for the target home appliances are obtained based on the parameter adaptation results; The device control unit is used to control target home appliances based on control commands.

[0084] In this embodiment, dimensionality analysis refers to structural analysis of the multidimensional data of the target scenario vector in order to separate and identify the sub-vectors representing different aspects of information.

[0085] In this embodiment, the multi-dimensional feature vector refers to a set of vectors obtained through dimensional analysis, where each vector represents a feature of a specific dimension (such as user status, environmental conditions, etc.) in the target scenario.

[0086] In this embodiment, the vector device mapping model refers to a pre-trained machine learning model whose function is to map or classify the input multi-dimensional feature vectors to one or more smart home devices that most need to be controlled.

[0087] In this embodiment, the parsing and identification result refers to the result output by the vector device mapping model after analyzing the input features, which usually includes the identification or type of the home device identified as the target.

[0088] In this embodiment, the control instruction element refers to the control instruction template or basic framework for a certain type of home appliance obtained from the system management terminal, which includes parameterizable parts.

[0089] In this embodiment, the motion control parameter refers to the numerical parameter (such as the percentage of light brightness or the target temperature value of the air conditioner) that is calculated or derived based on the current target scenario vector and is used to control the specific actions of the device.

[0090] In this embodiment, the execution instruction element refers to the configurable part of the control instruction element that specifically describes the device operation actions.

[0091] In this embodiment, parameter adaptation refers to the process of filling or replacing the specific values ​​of motion control parameters into the corresponding variable positions in the control instruction (especially its execution instruction).

[0092] In this embodiment, the parameter adaptation result refers to the final control command formed after parameter adaptation is completed, which contains specific control parameter values ​​and can be directly executed by the device.

[0093] The beneficial effects of the above technical solution are: by intelligently analyzing the characteristics of complex scenarios, accurately identifying the home devices that need to be controlled, and automatically generating control commands that are highly adapted to the current state, and by dynamically adjusting the execution parameters, ensuring that the device response perfectly matches the user's actual needs and environmental conditions, ultimately achieving personalized, precise, and automated smart home control, significantly improving the system's intelligence level and user experience.

[0094] Example 9: Based on Example 1, this example provides a smart home voice interaction control system based on a nursing bed. The control module includes: The data collection unit is used to acquire historical voice interaction control data between the user and home appliances within a target time period, as well as the scene information corresponding to each historical voice interaction control, and to extract the scene features corresponding to the scene information. Service strategy library building unit, used for: Based on scene features, the scene information corresponding to the historical voice interaction control data is classified into scenes, and scene recognition labels are generated for each category based on scene features. The scene classification results are then marked based on the scene recognition labels. Based on the labeling results, the historical voice interaction control data for each category of scenario is traversed, and the control time, control items and corresponding control quantification indicators for each historical voice interaction control data are determined based on the data traversal results. Construct a data control group for the same control project under the same category of scenarios, and determine the fluctuation range of the control quantitative indicators based on the data control group; Based on the fluctuation range, the action amplitude of each control item is limited. At the same time, based on the proportion of different control quantitative indicators in the fluctuation range, the user's control preference is determined. Based on the control time, action amplitude, and control preference, the user's control habits for each control item in different scenarios are obtained. A service strategy library is built based on users' control habits for various control items in different scenarios; Automatic service unit, used for: The system monitors the user's environment in real time and proactively sends service requests to the user when the time and environment meet the service policies already in the service policy library. Based on proactively sending results to monitor user feedback instructions, and upon receiving user feedback instructions, retrieve the target service policy corresponding to the current time node and scenario from the service policy library; Active control of home appliances based on target service strategies.

[0095] In this embodiment, historical voice interaction control data refers to the complete set of interaction records recorded by the system, which show that the user successfully controlled home appliances by voice within a certain period of time in the past.

[0096] In this embodiment, scene information refers to the set of background data related to the environment, user, and device status at each historical voice interaction control event.

[0097] In this embodiment, scene features refer to key parameters or vectors extracted from scene information that characterize the core attributes of the scene.

[0098] In this embodiment, scene classification refers to the process of categorizing different historical scene information into several categories based on the similarity of scene features.

[0099] In this embodiment, the scene identification label refers to the name or code assigned to each classified scene category to uniquely identify and distinguish the category.

[0100] In this embodiment, data traversal refers to systematically examining and analyzing all historical interaction data categorized under a specific scenario category.

[0101] In this embodiment, the control item refers to the specific function of the home device that the user intends to operate in voice interaction control (e.g., "adjust the air conditioner temperature" or "turn the main light on or off").

[0102] In this embodiment, the control quantification index refers to the specific, measurable numerical parameters involved when performing operations on the control item (e.g., temperature setpoint "26℃", brightness percentage "70%).

[0103] In this embodiment, the data control group refers to a data set formed by aggregating all historical control quantitative index values ​​for the same control item under the same scenario category for comparative analysis.

[0104] In this embodiment, the fluctuation range refers to the common range of quantitative indicator values ​​that users habitually set for the control item, determined by analyzing the data control group (e.g., the temperature is usually set between 24-28°C).

[0105] In this embodiment, the action limit refers to the safe execution range of parameters for automated control of the controlled item set by the system based on the calculated fluctuation range, in order to prevent abnormal operations that exceed the user's habits.

[0106] In this embodiment, control preference refers to the regular tendency of a user to choose a specific value or sub-range within a fluctuation range, which is derived from the analysis of the user's historical operation data.

[0107] In this embodiment, control habits refer to a stable behavioral pattern of how a user controls a device in a specific scenario, which is formed by combining the user's operation time pattern, the action limit set by the system, and the user's personal control preferences.

[0108] In this embodiment, the service policy library refers to a database that stores all learned rules or models about users' habits of controlling various devices in different scenarios.

[0109] In this embodiment, a service request refers to a prompt or inquiry that the system actively sends to the user when it detects that a certain service policy trigger condition is met, asking whether to execute a certain automated control.

[0110] In this embodiment, feedback instructions refer to the affirmative or negative response instructions given by the user to the service request actively issued by the system.

[0111] In this embodiment, the target service strategy refers to a personalized automated control scheme that is retrieved from the service strategy library, perfectly matches the current real-time time and scenario, and is about to be executed.

[0112] The beneficial effects of the above technical solution are as follows: by analyzing users' historical interaction data, it can autonomously learn control habits and preferences in different scenarios and build a personalized service strategy library accordingly. At the same time, it can perceive the user's scenario in real time, proactively initiate service inquiries at appropriate times, and automatically execute precise device control that conforms to the user's habits based on user feedback. This achieves a leap from passive response to proactive service, greatly improving the automation level and user experience of smart homes. Furthermore, it ensures the rationality and personalization of control behavior through quantitative indicator limits and preference recognition.

[0113] Example 10: This example provides a smart home voice interaction control method based on a nursing bed, such as... Figure 3 As shown, it includes: Step 1: Collect user vital signs data, environmental data, and bed posture data in real time; Step 2: Collect voice data in real time and perform noise reduction processing on the voice data to obtain standard voice data; Step 3: Based on user vital signs data, environmental data, bed posture data, and standard speech data, multimodal fusion is performed according to the time dimension to generate a target scenario vector; Step 4: Determine the target home appliances and control commands for the appliances based on the target scenario vector, and control the target home appliances according to the control commands.

[0114] The working principle and beneficial effects of the above technical solution are as follows: By collecting user vital signs data, environmental data, bed posture data, and voice data, and performing noise reduction processing on the collected voice data, accurate voice signals can be effectively determined. Secondly, the obtained user vital signs data, environmental data, bed posture data, and standard voice data are fused in a multimodal manner according to the time dimension to accurately and effectively determine the target scenario vector, thereby determining the user's comprehensive state at different times. Finally, based on the determined target scenario vector, the target home appliances and the control commands for the appliances are determined, ultimately enabling the control of the target home appliances according to the control commands. This provides great convenience for users, improves the effective integration of nursing beds and smart home appliances, and enhances the convenient interaction between nursing beds and smart home appliances.

[0115] Obviously, those skilled in the art can make various modifications and variations to this invention without departing from its spirit and scope. Therefore, if these modifications and variations fall within the scope of the claims of this invention and their equivalents, this invention also intends to include these modifications and variations.

Claims

1. A smart home voice interaction control system based on a nursing bed, characterized in that, include: The data acquisition module is used to collect user vital signs data, environmental data, and bed posture data in real time. The voice receiving module is used to collect voice data in real time and perform noise reduction processing on the voice data to obtain standard voice data. The multimodal fusion module is used to generate a target scenario vector by performing multimodal fusion based on user vital signs data, environmental data, bed posture data, and standard voice data according to the time dimension. The control module is used to determine the target home appliances and control commands for the appliances based on the target scenario vector, and to control the target home appliances according to the control commands.

2. The intelligent home voice interaction control system based on a nursing bed according to claim 1, characterized in that, The data acquisition module includes: The vital signs data acquisition unit is used to collect user vital signs data in real time based on the vital signs sensor array, and to perform initial storage of the user vital signs data; An environmental data acquisition unit is used to collect environmental data in real time from the area where the nursing bed is located based on environmental sensors, and to store the environmental data in a second location. The bed posture data acquisition unit is used to acquire linear displacement pulse signals in real time based on the sensing devices inside the nursing bed actuator, and to convert the acquired linear displacement pulse signals into posture data according to the data processing center integrated into the nursing bed to obtain bed posture data.

3. The intelligent home voice interaction control system based on a nursing bed according to claim 1, characterized in that, The voice reception and analysis module includes: The voice acquisition unit is used to acquire external voice data sequences in real time. The keyword matching unit is used to obtain preset voice keywords and match the preset voice keywords with external voice data sequences. When the preset voice keywords are present in the external voice data sequence in the matching result, a voice receiving command is generated. The voice acquisition unit is also used to take the time point corresponding to the preset voice keywords as the starting timestamp, and to acquire key voice data in real time according to the starting timestamp and the voice receiving instruction. The filtering unit is used to filter key speech data to obtain standard speech data.

4. A smart home voice interaction control system based on a nursing bed according to claim 3, characterized in that, The filtering unit includes: The audio acquisition subunit is used to acquire ambient noise audio from different sound sources in the nursing environment. The learning sub-unit is used to learn the environmental noise audio from different sound sources, determine the environmental noise audio characteristics of each sound source, and construct a noise recognition network based on the environmental noise audio characteristics of each sound source. The noise management strategy allocation subunit is used to acquire noise attributes, including steady-state noise and sudden speech interference noise, and to allocate different noise management strategies according to the noise attributes. The noise management mechanism determination subunit is used to associate the noise identification network with the noise management strategy to generate the noise management mechanism; The first processing subunit is used to input the key voice data collected in real time into the noise management mechanism for analysis, execute the corresponding noise management strategy according to the analysis results, and obtain the first audio information according to the execution results. The coherence determination subunit is used for: The first audio information is amplified and processed to obtain the second audio information, and the information content of the second audio information is identified and the coherence of the information content is judged. When the coherence of the information content reaches the preset standard, the second audio information will be output as the standard speech information. The second processing subunit is used to analyze the second audio information based on a preset semantic model when the coherence of the information content does not meet the preset standard, and to supplement the information content of the second audio information according to the analysis results until the supplemented information content meets the preset standard, and then use the supplemented second audio information as the standard speech information.

5. A smart home voice interaction control system based on a nursing bed according to claim 4, characterized in that, The first processing subunit includes: The identification subunit is used to input key speech data into the noise management mechanism for identification based on the noise identification network, and to determine the noise data and noise source of the key speech data. The strategy allocation subunit is used for: Obtain the noise attributes of noise sources and assign noise management strategies to denoise key speech data based on the noise attributes; When the noise source is a steady-state noise, the corresponding noise management strategy is to filter the noise data corresponding to the noise source based on a preset filtering model. When the noise source is characterized as sudden speech interference noise, the corresponding noise management strategy is to instantly reduce the gain of the air microphone to reduce the noise decibels corresponding to the noise data.

6. The intelligent home voice interaction control system based on a nursing bed according to claim 1, characterized in that, include: The smart nursing bed connects to home appliances based on a communication protocol and interfaces with the Internet through a preset gateway.

7. A smart home voice interaction control system based on a nursing bed according to claim 1, characterized in that, The multimodal fusion module includes: The data retrieval unit is used to retrieve the collected user vital signs data, environmental data, bed posture data, and standard voice data, and to extract the timestamps corresponding to the user vital signs data, environmental data, bed posture data, and standard voice data. Multimodal fusion unit, used for: Based on timestamps, user vital signs data, environmental data, bed posture data, and standard voice data under the same time dimension are time-series aligned, and a multimodal data stream with a unified timestamp is constructed based on the time-series alignment results. Feature extraction is performed on the multimodal data stream, and feature vectors for each data modality are obtained based on the feature extraction. The interaction relationships between multimodal data streams are determined based on a preset control protocol, and the feature vectors of each data modality are weighted and fused based on the interaction relationships and a preset attention mechanism. The target scenario vector is obtained based on the weighted fusion result.

8. A smart home voice interaction control system based on a nursing bed according to claim 1, characterized in that, The control module includes: Equipment determination unit, used for: The target scenario vector is obtained, and its dimensions are analyzed to determine the multi-dimensional feature vectors contained in the target scenario. The pre-trained vector device mapping model is used to analyze and identify multi-dimensional feature vectors, and the target home device is determined based on the analysis and identification results. Instruction generation unit, used for: The system obtains control command elements corresponding to the target home devices from the management terminal. At the same time, it determines the motion control parameters for each target home device based on the target scenario vector, and performs parameter adaptation on the execution command elements in the control command elements based on the motion control parameters. Control commands for the target home appliances are obtained based on the parameter adaptation results; The device control unit is used to control target home appliances based on control commands.

9. A smart home voice interaction control system based on a nursing bed according to claim 1, characterized in that, The control module includes: The data collection unit is used to acquire historical voice interaction control data between the user and home appliances within a target time period, as well as the scene information corresponding to each historical voice interaction control, and to extract the scene features corresponding to the scene information. Service strategy library building unit, used for: Based on scene features, the scene information corresponding to the historical voice interaction control data is classified into scenes, and scene recognition labels are generated for each category based on scene features. The scene classification results are then marked based on the scene recognition labels. Based on the labeling results, the historical voice interaction control data for each category of scenario is traversed, and the control time, control items and corresponding control quantification indicators for each historical voice interaction control data are determined based on the data traversal results. Construct a data control group for the same control project under the same category of scenarios, and determine the fluctuation range of the control quantitative indicators based on the data control group; Based on the fluctuation range, the action amplitude of each control item is limited. At the same time, based on the proportion of different control quantitative indicators in the fluctuation range, the user's control preference is determined. Based on the control time, action amplitude, and control preference, the user's control habits for each control item in different scenarios are obtained. A service strategy library is built based on users' control habits for various control items in different scenarios; Automatic service unit, used for: The system monitors the user's environment in real time and proactively sends service requests to the user when the time and environment meet the service policies already in the service policy library. Based on proactively sending results to monitor user feedback instructions, and upon receiving user feedback instructions, retrieve the target service policy corresponding to the current time node and scenario from the service policy library; Active control of home appliances based on target service strategies.

10. A smart home voice interaction control method based on a nursing bed, characterized in that, include: Step 1: Collect user vital signs data, environmental data, and bed posture data in real time; Step 2: Collect voice data in real time and perform noise reduction processing on the voice data to obtain standard voice data; Step 3: Based on user vital signs data, environmental data, bed posture data, and standard speech data, multimodal fusion is performed according to the time dimension to generate a target scenario vector; Step 4: Determine the target home appliances and control commands for the appliances based on the target scenario vector, and control the target home appliances according to the control commands.

Images