Intelligent eye-protection illumination sitting posture correction method and system

Abstract

The invention discloses an intelligent eye-protection illumination sitting posture correction method and system, particularly relates to the technical field of illumination sitting posture correction, and is used for solving the problem of asynchronous illumination sitting posture analysis. According to the method, data from different sensors are ensured to be synchronized on the same time reference through a dynamic time warping algorithm, errors caused by asynchronization are eliminated, a self-attention mechanism is combined with position coding, the complex relation and time sequence dependence of the data are accurately extracted, the feature representation capability of the model is improved, and finally, the model is optimized. Time sequence data fusion is carried out on the multi-modal features according to the long-short-term memory network, the user state and demand are predicted in real time, accurate light adjustment and sitting posture monitoring are achieved, the robustness and adaptability of data analysis are improved, the user experience and health management efficiency can be remarkably improved in intelligent eye-protection illumination sitting posture correction, and the user experience is improved. And a more comfortable and healthy working and living environment is provided for the user.

Description

Technical Field

[0001] The present invention relates to the technical field of lighting sitting posture correction, and more specifically, to an intelligent eye protection lighting sitting posture correction method and system. Background Art

[0002] With the advancement of science and technology and the improvement of living standards, people's working, learning and living environments have undergone tremendous changes. In particular, the widespread use of electronic devices has made eye health issues increasingly concerned. At the same time, the use of electronic devices is often accompanied by poor sitting posture, which not only affects spinal health, but may also lead to muscle strain, poor blood circulation and other health problems. Especially in modern society where people sit for long periods of time to work and study, sitting posture problems are particularly prominent.

[0003] Deficiencies of existing technologies: With the popularization of electronic devices and long-term sedentary work environments, the widespread use of these devices has improved efficiency and convenience, but it has also brought significant health challenges, especially the impact on eyes and body posture. Eye health and sitting posture problems have become the main factors affecting the health of modern people. The eye protection lighting and sitting posture correction technologies on the current market mainly rely on independent hardware devices and simple monitoring methods. Although these technologies can alleviate eye and body fatigue to a certain extent, the solutions are often single-function and lack comprehensive consideration of the relationship between lighting and sitting posture correction. Summary of the invention

[0004] In order to overcome the above-mentioned defects of the prior art, the embodiments of the present invention provide an intelligent eye protection lighting sitting posture correction method and system. The present invention ensures that the data from different sensors are synchronized on the same time basis through a dynamic time warping algorithm, eliminates the errors caused by asynchrony, and then combines the self-attention mechanism with position encoding to accurately extract the complex relationships and timing dependencies of the data, thereby improving the feature representation capability of the model. Finally, the multimodal features are fused with time series data based on the long short-term memory network, and the user status and needs are predicted in real time to achieve accurate light adjustment and sitting posture monitoring, so as to solve the problems raised in the above-mentioned background technology.

[0005] To achieve the above object, the present invention provides the following technical solutions: The intelligent eye protection lighting sitting posture correction method comprises the following steps: Acquire sensor data for monitoring and pre-process the data collected by each sensor; Mark the pre-processed sensor data according to the timestamp, use the dynamic time warping algorithm to synchronize the data, and determine the time reference of the data; For data with the same time base, the self-attention mechanism is used to determine the relationship and temporal dependency between the data, and the multimodal features of the data are extracted in combination with position encoding; The extracted multimodal features are input into the long short-term memory network for time series data fusion and training to predict the user's status and needs. The trained model is used to analyze the sensor data in real time and output optimized light adjustment and sitting posture monitoring suggestions.

[0006] In a preferred embodiment, the sensor data for monitoring is obtained and the data collected by each sensor is preprocessed. The specific process is as follows: The sensors include light sensors, environmental sensors, pressure sensors and posture sensors. The environmental sensors include temperature and humidity sensors, and the posture sensors include accelerometers and gyroscopes. Perform data preprocessing on the collected sensor data, including filtering, denoising and normalization; Kalman filtering and LMS algorithm are used to eliminate data noise and environmental interference.

[0007] In a preferred embodiment, Kalman filtering and LMS algorithm are used to eliminate data noise and environmental interference. The specific process is as follows: De-noise the data, convert the sensor data into signal form, set the initial state estimate and initial covariance matrix, predict the state value and covariance matrix from k to k-1 at the next moment according to the state transition model of Kalman filtering, update the state estimate and covariance matrix according to the observed value, update the state of the prediction result, and determine the Kalman gain matrix; Combine the observations and the Kalman gain matrix to update the state estimate and covariance at the current moment, process the observation data at each time step, and output the smoothed signal; The LMS algorithm calculates the filter output and the filter error, updates the filter weight according to the filter error, and adjusts the filter weight for all data filter outputs and filter errors.

[0008] In a preferred embodiment, a dynamic time warping algorithm is used to synchronize data and determine the time reference of the data. The specific process is as follows: Calculate the Euclidean distance matrix between time series; Construct a cumulative distance matrix and take the path with the minimum cumulative distance from the end point back to the starting point as the optimal matching path; The best matching path is used as the end point value of the cumulative distance matrix; Compare endpoint values ​​and align the time base of unsynchronized sensor data.

[0009] In a preferred embodiment, the self-attention mechanism is used to determine the relationship and temporal dependency between data, and the multimodal features of the data are extracted in combination with position encoding. The specific process is as follows: Define sensor data as an input sequence and build query, key, and value matrices; Use the self-attention mechanism to calculate the attention score of each element in the input sequence, and determine multi-level information through parallel calculation of multi-head self-attention mechanisms; Add position encoding to the input sequence to determine the corresponding timing information, combine the attention mechanism output and position encoding, and extract the multimodal features of the data.

[0010] In a preferred embodiment, the trained model is used to analyze sensor data in real time to output optimized light adjustment and sitting posture monitoring suggestions. The specific process is as follows: Based on the fused data and prediction results, combined with the ambient light intensity, user activity status and time information, the light settings are optimized; The data from the pressure sensor and posture sensor, combined with the output of the LSTM model, provide suggestions for posture correction and remind users to adjust their sitting posture through vibration reminders, sound prompts or visual feedback.

[0011] The intelligent eye protection lighting sitting posture correction system is used for the above-mentioned intelligent eye protection lighting sitting posture correction method, comprising: A data processing module, used to obtain sensor data for monitoring and pre-process the data collected by each sensor; Dynamic synchronization module, used to mark the pre-processed sensor data according to the timestamp, synchronize the data using the dynamic time warping algorithm, and determine the time reference of the data; The multimodal feature extraction module is used to determine the relationship and temporal dependency between data with the same time base using the self-attention mechanism, and extract the multimodal features of the data in combination with position encoding; The analysis output module is used to input the extracted multimodal features into the long short-term memory network for time series data fusion and training, predict the user's status and needs, use the trained model to analyze the sensor data in real time, and output optimized light adjustment and sitting posture monitoring suggestions.

[0012] Technical effects and advantages of the present invention: Through the dynamic time warping algorithm, we ensure that data from different sensors are synchronized on the same time basis to eliminate the errors caused by asynchrony. Then, we combine the self-attention mechanism with position encoding to accurately extract the complex relationships and temporal dependencies of the data and improve the feature representation ability of the model. Finally, we use the long short-term memory network to fuse multimodal features into time series data, predict user status and needs in real time, and achieve accurate light adjustment and sitting posture monitoring.

[0013] Through the combination of self-attention mechanism and position encoding, it is possible to accurately capture the complex relationships in multimodal data, improve the accuracy of model predictions, and comprehensively consider the relationship between lighting and posture correction. Through multimodal data fusion, a comprehensive health management solution is provided. The dynamic time warping algorithm ensures data synchronization, and can effectively align data even in the case of asynchronous or unequal length data, thereby improving the robustness and adaptability of data analysis. In the intelligent eye protection lighting posture correction, it can significantly improve the user experience and the efficiency of health management, providing users with a more comfortable and healthy working and living environment. BRIEF DESCRIPTION OF THE DRAWINGS

[0014] Figure 1 The present invention is a flow chart of the intelligent eye protection lighting sitting posture correction method.

[0015] Figure 2 It is a structural schematic diagram of the intelligent eye protection lighting sitting posture correction system of the present invention. DETAILED DESCRIPTION

[0016] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0017] Example 1

[0018] like Figure 1 As shown, the intelligent eye protection lighting sitting posture correction method comprises: Acquire sensor data for monitoring and pre-process the data collected by each sensor; Mark the pre-processed sensor data according to the timestamp, use the dynamic time warping algorithm to synchronize the data, and determine the time reference of the data; For data with the same time base, the self-attention mechanism is used to determine the relationship and temporal dependency between the data, and the multimodal features of the data are extracted in combination with position encoding; The extracted multimodal features are input into the long short-term memory network for time series data fusion and training to predict the user's status and needs. The trained model is used to analyze the sensor data in real time and output optimized light adjustment and sitting posture monitoring suggestions.

[0019] A multifunctional intelligent chair integrates a light sensor, an environmental sensor, a pressure sensor and a posture sensor. The integrated light sensor is installed on the desktop to detect the intensity and changes of ambient light. The environmental sensor includes a temperature and humidity sensor installed on the side or bottom of the table to detect ambient temperature and humidity. The pressure sensor is embedded in the seat cushion and backrest of the chair to detect the user's sitting posture and weight distribution. The posture sensor includes an accelerometer and a gyroscope, which are installed on the chair backrest and seat cushion to monitor the user's posture changes in real time. Smart lighting equipment includes full-spectrum LED lights, which provide a spectrum close to natural light and reduce eye fatigue. The lamps are designed to be adjustable in angle and height to adapt to different usage scenarios; The brightness and color temperature regulators are integrated inside the lamps to dynamically adjust the brightness and color temperature through the light sensor data. The wireless communication module is also integrated to support wireless connection with smart tables and chairs and control systems to achieve data transmission and device control. At the same time, it has biofeedback devices, such as smart watches or bracelets, which are equipped with heart rate monitoring, galvanic skin response sensors, etc. to monitor the user's physiological signals in real time, or non-contact physiological monitoring sensors installed near tables and chairs, which use infrared or radar technology to monitor the user's heart rate, breathing rate and other data; Through multi-sensor data fusion, intelligent control of ambient light and sitting posture is achieved, sensor data is analyzed, lighting brightness and color temperature are adjusted in real time, and user sitting posture is monitored and corrected.

[0020] The collected sensor data is preprocessed to filter, denoise and normalize to ensure data quality. Based on the machine learning model, the brightness and color temperature of the light source are adjusted in real time by analyzing the ambient light and user activity data. The data of the pressure sensor and posture monitoring sensor are analyzed in real time to detect bad sitting posture and provide correction suggestions. Finally, the data collected by each sensor in real time is transmitted to the central processing unit through wireless communication. The central processing unit fuses and analyzes the sensor data to generate real-time lighting adjustment and sitting posture suggestions. The specific steps are as follows: Pre-process the data collected by each sensor, apply Kalman filtering and adaptive filtering algorithms to eliminate noise and environmental interference, and ensure the accuracy and stability of the data; Setting the initial state estimate and the initial covariance matrix and converting sensor data into signal form; According to the state transition model of Kalman filtering, the state value and covariance matrix from the next moment k to k-1 are predicted. The expression of the estimated value and covariance matrix in the state prediction process is: , , where A represents the state transfer matrix, represents the transpose of the matrix, B represents the control matrix, represents the control input, Q represents the process noise covariance matrix; Update the state estimate and covariance matrix according to the observed value, that is, update the state of the prediction result and determine the Kalman gain matrix: , status update, , covariance update, , where H represents the observation matrix, R represents the observation noise covariance matrix, represents the Kalman gain matrix, Represents the observation value at the kth moment, I represents the unit matrix, and then repeats the above steps to process the observation data of each time step, that is, to process the sensor data signal at each time and output a smoother and more accurate signal.

[0021] Adaptive filtering is a technology that dynamically adjusts filter parameters according to changes in the input signal. This implementation uses the LMS (Least Mean Squares) algorithm to calculate the filter output: , calculate the filtering error, the calculation expression is: , update the weights: ,in, is the i-th filter weight, μ is the step size factor, e(n) is the error signal, d(n) is the desired signal, x(n) represents the input signal, and n is the signal data sample; It should be noted that the state transition model is a mathematical model used to describe the change of data state over time. It is widely used in signal processing, control systems, time series analysis and other fields. It predicts the future state of data by describing the state of data at different times and its transfer relationship. Kalman filtering is an algorithm that uses the state transition model for prediction and update.

[0022] Adaptive filtering and noise elimination algorithms can effectively reduce noise in sensor data and provide more accurate ambient light, posture and physiological data. Through Kalman filtering and adaptive filtering, data can be processed and updated in real time to ensure fast response and adjustment, filter out noise and dynamically adjust filter parameters, thereby maintaining stable and reliable performance under various environmental conditions. High-quality data processing can provide more accurate light adjustment and sitting posture correction suggestions, thereby improving user comfort and health.

[0023] For example, Kalman filtering is used to process light sensor data to eliminate environmental noise, ensure that lighting equipment adjusts brightness and color temperature based on accurate light data, and apply adaptive filtering to process environmental sensor data to ensure the stability of environmental parameters (such as temperature and humidity) and optimize light adjustment strategies. Kalman filtering is used to process pressure sensor and posture sensor data to monitor the user's sitting posture in real time and provide accurate sitting status. Adaptive filtering is used to process biofeedback data (such as heart rate and breathing rate) to detect the user's physiological state in real time and provide comprehensive health advice based on sitting posture data.

[0024] Use timestamps to mark pre-processed multi-sensor data, and use the dynamic time warping (DTW) algorithm to synchronize data, ensuring that multi-sensor data is processed on the same time basis, avoiding analysis errors caused by data asynchrony, and ensuring response speed and accuracy; Dynamic time warping is used to synchronize the time series of multi-sensor data. Dynamic time warping is an algorithm used to measure the similarity between two time series signals. It is particularly suitable for signals of different lengths or with time series deviations. For two given time series: , calculate the Euclidean distance matrix between time series: ; Construct a cumulative distance matrix C, where C(i, j) represents the minimum cumulative distance from the starting point to (i, j): , where the initial condition is C(1,1)=D(1,1), and the boundary conditions are: C(i,1)=D(i,1)+C(−1,1), C(1,j)=D(1,j)+C(1,j−1), where the value of i is in the range of 2-n, and the value of j is in the range of 2-m; From the end point (n, m) back to the starting point (1, 1), find the path with the minimum cumulative distance. satisfy , the final DTW distance is the end point value of the cumulative distance matrix; Calculate the Euclidean distance between each point in the time series X and Y of the data to form a matrix D. Use the recursive formula to fill in the cumulative distance matrix C. Backtrack from the end point of the cumulative distance matrix to find the optimal matching path. The final DTW distance is the end point value of the cumulative distance matrix. The optimal matching path is used to align the time base of the asynchronous sensor data. For each sequence, the timestamp is adjusted according to the optimal matching path so that the data are aligned on the same time base.

[0025] By comparing endpoint values, similar patterns and behaviors in different time series can be identified. For example, in eye protection lighting and sitting posture correction, the user's behavior patterns in different time periods are analyzed; the endpoint value provides data support for the decision-making system to adjust the light and sitting posture. For example, if the behavior in two time periods is similar, the system can optimize the current control plan based on the historical adjustment strategy.

[0026] Aligning asynchronous sensor data enables accurate analysis and decision-making based on the aligned data. By aligning the time base of the sensor data, it can more accurately monitor the ambient light and user sitting status, improve the accuracy and reliability of regulation, and process data of different lengths or time deviations, thereby improving adaptability in complex and dynamic environments.

[0027] Perform multimodal feature extraction on the data after synchronization of the time base, and capture the complex relationship and temporal dependency between the data through the self-attention mechanism; Representing multimodal sensor data as an input sequence , n is a positive integer; Create query, key and value matrices. The query matrix Q represents the input at the current moment. , the key matrix represents the input at all times, , the value matrix represents the input at all times, ,in, , , represents a trainable weight matrix that maps inputs to query, key, and value spaces; The self-attention mechanism is used to calculate the attention score of each element in the input sequence. , where Attention(Q, K, V) is the attention output, representing the weighted value matrix, and softmax is the normalization function used to convert the score into a probability distribution. is the dot product of the query and the key, indicating the similarity between the query and the key, represents the scaling factor, balancing the magnitude of the dot product, is the dimension of the key; Capture multi-level information through parallel calculation of multi-head self-attention mechanism, and calculate for each head: , represents the weight matrix of the i-th head; The input sequence adds position coding to capture timing information. The position coding expression is: , , where PE is the position code, pos is the position index, and i is the dimension index. is the model dimension; The context vector calculated by the self-attention mechanism is combined with the position encoding to form a multimodal feature representation containing temporal information; Through the self-attention mechanism and position encoding, it is possible to effectively extract multimodal features of the data and capture the complex relationships between temporal dependencies and data. These features can be used for subsequent temporal data fusion and state prediction, improving the accuracy and responsiveness of the intelligent eye protection lighting and sitting posture correction system.

[0028] The extracted multimodal features are input into the long short-term memory network (LSTM) for time series data fusion, predicting the user's status and needs. The trained model is used to analyze sensor data in real time, output optimized light adjustment and sitting posture monitoring suggestions, and through the deep learning model, comprehensive analysis and intelligent control of multi-sensor data are achieved to provide more accurate light and sitting posture management.

[0029] Based on the fused data and prediction results, the brightness and color temperature of the lighting equipment are dynamically adjusted. The light settings are optimized based on the ambient light intensity, user activity status and time information. The user's sitting posture is monitored in real time. Through the data of the pressure sensor and posture sensor, combined with the output of the LSTM model, sitting posture correction suggestions are provided, and the user is reminded to adjust his sitting posture through vibration reminders, sound prompts or visual feedback. Based on the fused data and prediction results, the brightness and color temperature of the lighting equipment are dynamically adjusted. The brightness adjustment process is expressed as follows: brightness = f (ambient light intensity, user activity status). Ambient light intensity is obtained from the light sensor, and user activity status is obtained through LSTM model prediction. f represents the brightness adjustment function. Color temperature = g(time, ambient light color temperature, user preference), where time is the current time, ambient light color temperature is obtained from the light sensor, and user preference is the color temperature preference set by the user; According to the output of the LSTM model, the user's sitting posture is monitored and corrected in real time. The sitting posture state = h (pressure distribution, posture angle). The pressure distribution is obtained from the pressure sensor, the posture angle is obtained from the posture sensor, and h represents the sitting posture adjustment function. The color temperature adjustment function g is a function used to dynamically adjust the color temperature of lighting equipment according to different input parameters. Specifically, g represents the following: time represents the current time point, taking into account the requirements for color temperature in different time periods of the day. For example, warmer color temperatures are required in the morning and evening, while cooler color temperatures are suitable during the day; ambient light color temperature represents the color temperature of the current ambient light, which is obtained through a light sensor. This parameter helps adjust the indoor lighting color temperature to be consistent with the ambient light, providing a more comfortable visual experience; user preference represents the user's personalized color temperature preference, which is obtained through user settings or long-term usage data analysis. This parameter ensures that the system meets the user's personalized needs and provides customized lighting solutions; The color temperature adjustment function g combines the above parameters to calculate and output a color temperature value suitable for the current situation. The specific calculation logic can be adjusted according to actual needs and design; For example, the current time is 14:00 (2pm), the ambient light color temperature is 5000K (detected by the light sensor), and the user prefers a cool color temperature, set to 5500K, the target color temperature is calculated based on the time period, ambient light color temperature, and user preference; The color temperature range in the morning (6:00-9:00) is 3000K-3500K, the color temperature range in the daytime (9:00-18:00) is 5000K-6000K, the color temperature range in the evening (18:00-21:00) is 3000K-3500K, and the color temperature range at night (21:00-6:00) is 2500K-3000K. It is adjusted appropriately according to the current ambient light color temperature and user preferences; Determine the basic color temperature range. The current time is 14:00, which is daytime (9:00-18:00), and the color temperature range is 5000K-6000K. Combined with the current ambient light color temperature of 5000K, it is within the basic color temperature range and does not require additional adjustment. The user preference is 5500K. Within the basic color temperature range, the final calculated target color temperature is: T=3 / (5000+6000+5500)=5500K. Based on the calculation results, the intelligent lighting system adjusts the color temperature of the indoor lighting to 5500K to meet user needs and environmental conditions.

[0030] Similarly, the brightness adjustment function f, the sitting posture adjustment function h, and the color temperature adjustment function g are set in the same way.

[0031] Furthermore, an intuitive user interface is provided in the user interface, and users can set personalized lighting and sitting posture management plans through the application, view health reports and receive adjustment suggestions. It can also support integration with mainstream voice assistants, and users can control lighting and receive sitting posture reminders through voice commands, that is, reminders can be issued through mobile applications or voice assistants, suggesting that users adjust their postures.

[0032] Subsequently, based on the user's physiological signals (such as heart rate and skin galvanic response) monitored by the biofeedback device, the lighting and sitting posture suggestions are optimized in real time to provide a healthier environment. The system continuously monitors the user's environmental and physiological data, dynamically adjusts the light source and sitting posture reminders to ensure that the user is always in the best condition; In application scenarios, such as working environment, smart tables and chairs and smart lighting equipment are set up in the office, and biofeedback equipment is installed. Users can set personalized health management plans through mobile applications or voice assistants. The system monitors and adjusts light sources and sitting postures in real time, and provides health reminders and suggestions; By comprehensively considering lighting and posture correction feedback, we provide comprehensive health support, effectively prevent eye and body fatigue, and improve users' work, study and life efficiency through personalized adjustment and management, providing a more comfortable user experience. Through data collection and analysis, we continuously optimize system performance and user experience, and provide more accurate health recommendations and adjustment plans.

[0033] The present invention uses a dynamic time warping algorithm to ensure that data from different sensors are synchronized on the same time basis, eliminating errors caused by asynchrony. It then combines the self-attention mechanism with position encoding to accurately extract the complex relationships and timing dependencies of the data, thereby improving the feature representation capability of the model. Finally, it performs time series data fusion on multimodal features based on the long short-term memory network, predicts user status and needs in real time, and achieves accurate light adjustment and sitting posture monitoring.

[0034] Through the combination of self-attention mechanism and position encoding, it is possible to accurately capture the complex relationships in multimodal data, improve the accuracy of model predictions, and comprehensively consider the relationship between lighting and posture correction. Through multimodal data fusion, a comprehensive health management solution is provided. The dynamic time warping algorithm ensures data synchronization, and can effectively align data even in the case of asynchronous or unequal length data, thereby improving the robustness and adaptability of data analysis. In the intelligent eye protection lighting posture correction, it can significantly improve the user experience and the efficiency of health management, providing users with a more comfortable and healthy working and living environment.

[0035] Embodiment 2, this embodiment is a system embodiment of embodiment 1, which is used to implement the intelligent eye protection lighting sitting posture correction method introduced in embodiment 1, such as Figure 2 As shown, specifically including: A data processing module, used to obtain sensor data for monitoring and pre-process the data collected by each sensor; Dynamic synchronization module, used to mark the pre-processed sensor data according to the timestamp, synchronize the data using the dynamic time warping algorithm, and determine the time reference of the data; The multimodal feature extraction module is used to determine the relationship and temporal dependency between data with the same time base using the self-attention mechanism, and extract the multimodal features of the data in combination with position encoding; The analysis output module is used to input the extracted multimodal features into the long short-term memory network for time series data fusion and training, predict the user's status and needs, use the trained model to analyze the sensor data in real time, and output optimized light adjustment and sitting posture monitoring suggestions.

[0036] The above formulas are all dimensionless and calculated numerically. Specific dimension removal can be achieved by various means such as standardization, which will not be elaborated here. The formula is a formula for the most recent real situation obtained by collecting a large amount of data and performing software simulation. The preset parameters in the formula are set by technicians in this field according to actual conditions.

[0037] The above embodiments may be implemented in whole or in part by software, hardware, firmware or any other combination thereof. When implemented by software, the above embodiments may be implemented in whole or in part in the form of a computer program product. The computer program product includes one or more computer instructions or computer programs. When the computer instructions or computer programs are loaded or executed on a computer, the process or function described in the embodiment of the present application is generated in whole or in part. The computer may be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions may be stored in a computer-readable storage medium, or transmitted from one computer-readable storage medium to another computer-readable storage medium. For example, the computer instructions may be transmitted from one website, computer, server or data center to another website, computer, server or data center by wired or wireless (e.g., infrared, wireless, microwave, etc.). The computer-readable storage medium may be any available medium that a computer can access or a data storage device such as a server or data center that contains one or more available media sets. The available medium may be a magnetic medium (e.g., a floppy disk, an ATA hard disk, a tape), an optical medium (e.g., a DVD), or a semiconductor medium. The semiconductor medium may be a solid-state ATA hard disk.

[0038] It should be understood that in the various embodiments of the present application, the size of the serial numbers of the above-mentioned processes does not mean the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present application.

[0039] Those of ordinary skill in the art will appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of this application.

[0040] In the several embodiments provided in the present application, it should be understood that the disclosed systems, devices and methods can be implemented in other ways. For example, the device embodiments described above are only schematic. For example, the division of the units is only a logical function division. There may be other division methods in actual implementation, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of devices or units, which can be electrical, mechanical or other forms.

[0041] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, and may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0042] In addition, each functional unit in each embodiment of the present application may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit.

[0043] The above is only a specific implementation of the present application, but the protection scope of the present application is not limited thereto. Any person skilled in the art who is familiar with the present technical field can easily think of changes or substitutions within the technical scope disclosed in the present application, which should be included in the protection scope of the present application. Therefore, the protection scope of the present application should be based on the protection scope of the claims.

Claims

1. Intelligent eye protection lighting sitting posture correction method, characterized in that: The steps include: Acquire sensor data for monitoring and pre-process the data collected by each sensor; Mark the pre-processed sensor data according to the timestamp, use the dynamic time warping algorithm to synchronize the data, and determine the time reference of the data; For data with the same time base, the self-attention mechanism is used to determine the relationship and temporal dependency between the data, and the multimodal features of the data are extracted in combination with position encoding; The extracted multimodal features are input into the long short-term memory network for time series data fusion and training to predict the user's status and needs. The trained model is used to analyze the sensor data in real time and output optimized light adjustment and sitting posture monitoring suggestions.

2. The intelligent eye protection lighting sitting posture correction method according to claim 1 is characterized in that: Obtain sensor data for monitoring and preprocess the data collected by each sensor. The specific process is as follows: The sensors include light sensors, environmental sensors, pressure sensors and posture sensors. The environmental sensors include temperature and humidity sensors, and the posture sensors include accelerometers and gyroscopes. Perform data preprocessing on the collected sensor data, including filtering, denoising and normalization; Kalman filtering and LMS algorithm are used to eliminate data noise and environmental interference.

3. The intelligent eye protection lighting sitting posture correction method according to claim 2 is characterized in that: Kalman filtering and LMS algorithm are used to eliminate data noise and environmental interference. The specific process is as follows: De-noise the data, convert the sensor data into signal form, set the initial state estimate and initial covariance matrix, predict the state value and covariance matrix from k to k-1 at the next moment according to the state transition model of Kalman filtering, update the state estimate and covariance matrix according to the observed value, update the state of the prediction result, and determine the Kalman gain matrix; Combine the observations and the Kalman gain matrix to update the state estimate and covariance at the current moment, process the observation data at each time step, and output the smoothed signal; The LMS algorithm calculates the filter output and the filter error, updates the filter weight according to the filter error, and adjusts the filter weight for all data filter outputs and filter errors.

4. The intelligent eye protection lighting sitting posture correction method according to claim 3 is characterized in that: Use the dynamic time warping algorithm to synchronize data and determine the time base of the data. The specific process is as follows: Calculate the Euclidean distance matrix between time series; Construct a cumulative distance matrix and take the path with the minimum cumulative distance from the end point back to the starting point as the optimal matching path; The best matching path is used as the end point value of the cumulative distance matrix; Compare endpoint values ​​and align the time base of unsynchronized sensor data.

5. The intelligent eye protection lighting sitting posture correction method according to claim 4 is characterized in that: The self-attention mechanism is used to determine the relationship and temporal dependency between data, and the multimodal features of data are extracted in combination with position encoding. The specific process is as follows: Define sensor data as an input sequence and build query, key, and value matrices; Use the self-attention mechanism to calculate the attention score of each element in the input sequence, and determine multi-level information through parallel calculation of multi-head self-attention mechanisms; Add position encoding to the input sequence to determine the corresponding timing information, combine the attention mechanism output and position encoding, and extract the multimodal features of the data.

6. The intelligent eye protection lighting sitting posture correction method according to claim 5 is characterized in that: Use the trained model to analyze sensor data in real time and output optimized light adjustment and sitting posture monitoring suggestions. The specific process is as follows: Based on the fused data and prediction results, combined with the ambient light intensity, user activity status and time information, the light settings are optimized; The data from the pressure sensor and posture sensor, combined with the output of the LSTM model, provide suggestions for posture correction and remind users to adjust their sitting posture through vibration reminders, sound prompts or visual feedback.

7. An intelligent eye-protection lighting sitting posture correction system, used to implement the intelligent eye-protection lighting sitting posture correction method according to any one of claims 1 to 6, characterized in that: include: A data processing module, used to obtain sensor data for monitoring and pre-process the data collected by each sensor; Dynamic synchronization module, used to mark the pre-processed sensor data according to the timestamp, synchronize the data using the dynamic time warping algorithm, and determine the time reference of the data; The multimodal feature extraction module is used to determine the relationship and temporal dependency between data with the same time base using the self-attention mechanism, and extract the multimodal features of the data in combination with position encoding; The analysis output module is used to input the extracted multimodal features into the long short-term memory network for time series data fusion and training, predict the user's status and needs, use the trained model to analyze the sensor data in real time, and output optimized light adjustment and sitting posture monitoring suggestions.

Images