An intelligent humidity real-time monitoring system based on bionic flexible sensing

Abstract

This invention proposes an intelligent real-time humidity monitoring system based on biomimetic flexible sensing, comprising a biomimetic flexible sensing module, an algorithm processing module, a model prediction module, and an intelligent terminal interaction module. The biomimetic flexible sensing module acquires visualized humidity response image signals through a double-layer flexible structure, which are transmitted in real-time to the algorithm processing module. The algorithm processing module encompasses image preprocessing, feature extraction, noise filtering, and dataset construction. A curvature recognition algorithm, combined with digital image processing, accurately extracts multi-dimensional features of surface curvature and texture distribution, constructing over 800 sets of multi-gradient humidity samples to solve the problems of nonlinear signal noise interference and feature mining in flexible sensors. The model prediction module achieves accurate humidity prediction based on a Python machine learning regression model, with an average absolute error as low as 1.8%RH and a single-frame inference time of <35ms. The intelligent terminal interaction module uses a Flask backend and a visual interface to achieve real-time sensor data access, online humidity prediction, and dynamic display of results, supporting real-time user monitoring and engineering implementation.

Description

Technical Field

[0001] This invention relates to the field of intelligent sensing and environmental monitoring, and in particular to an intelligent real-time humidity monitoring system based on biomimetic flexible sensing, which is suitable for accurate humidity monitoring in various scenarios such as industrial environment, agricultural planting, and smart home. Background Technology

[0002] Humidity monitoring is a core component of environmental perception and intelligent control. Traditional humidity sensors are mostly rigid structures, which have problems such as poor adaptability and weak surface fit. While biomimetic flexible double-layer sensors have the advantages of flexible fit and high sensitivity, their output signals have significant nonlinear characteristics and are easily affected by environmental interference, resulting in noise and insufficient humidity detection accuracy.

[0003] Existing humidity monitoring systems largely rely on single signal acquisition and simple algorithm processing, lacking a dedicated data processing system for the nonlinear characteristics of flexible sensors. This makes it difficult to effectively extract the core features of the sensor response, and they suffer from large model prediction errors and slow inference speeds, failing to meet real-time monitoring needs. Furthermore, traditional monitoring systems have rudimentary interfaces and low data visualization capabilities, unable to achieve real-time access to sensor data and dynamic display of monitoring results, resulting in poor engineering feasibility. With the development of flexible sensing technology and machine learning, the research and development of dedicated algorithms and integrated monitoring systems for biomimetic flexible sensors has become a trend. However, a complete humidity monitoring system covering data acquisition, algorithm analysis, model prediction, and system development has not yet been established, making it difficult to fully leverage the technological advantages of biomimetic flexible sensors. Summary of the Invention

[0004] The purpose of this invention is to address the challenges of nonlinear signal processing, low humidity monitoring accuracy, poor real-time performance, and insufficient system engineering application of biomimetic flexible double-layer sensors. This invention proposes an intelligent real-time humidity monitoring system based on biomimetic flexible sensors. It establishes a complete system for humidity data acquisition and algorithm analysis, using a self-designed curvature recognition algorithm combined with digital image algorithms to refine the visualization of sensor response images, extract multi-dimensional features, and construct a multi-gradient humidity sample dataset. Combined with a machine learning regression model, it achieves accurate humidity prediction. Simultaneously, an integrated front-end and back-end system is developed to enable real-time sensor data access, online humidity prediction, and dynamic display of results, thereby improving the accuracy, real-time performance, and engineering application capabilities of humidity monitoring.

[0005] The technical implementation scheme of the present invention is as follows: the intelligent humidity real-time monitoring system based on biomimetic flexible sensing includes a biomimetic flexible sensing module, a data acquisition module, an algorithm processing module, a model prediction module, and an intelligent terminal interaction module; the algorithm processing module includes core processes such as image preprocessing, feature extraction, and noise filtering, and the system realizes integrated management and control of algorithm design, data preprocessing, model iterative optimization, and engineering implementation throughout the entire process.

[0006] Furthermore, the biomimetic flexible sensing module adopts a biomimetic flexible dual-layer sensor structure, which has flexible curved surface fitting characteristics and high humidity response sensitivity. It can be adapted to monitoring scenarios of different forms and outputs a visualized image signal of humidity response, providing a foundation for subsequent data processing.

[0007] Furthermore, the data acquisition module enables real-time acquisition of the response signal of the biomimetic flexible sensor, synchronously acquires sensor visualization image data, constructs a data transmission channel, and transmits the raw data to the algorithm processing module without delay, ensuring the integrity and real-time performance of the data acquisition.

[0008] Furthermore, the algorithm processing module first uses traditional digital image algorithms to preprocess the sensor response visualization image, sequentially completing grayscale normalization, adaptive threshold segmentation, and morphological filtering to effectively eliminate image noise and improve image quality.

[0009] Furthermore, the algorithm processing module independently designs a curvature recognition algorithm, and combined with the image preprocessing results, innovatively extracts multi-dimensional features such as surface curvature and texture distribution of the sensor response image, and mines core feature parameters that are strongly correlated with humidity, providing high-quality feature input for model training.

[0010] Furthermore, the algorithm processing module constructs a multi-gradient humidity sample dataset of 800+ groups based on multi-dimensional feature parameters. Through feature selection and precise noise filtering, invalid data and interfering features are eliminated, the dataset quality is optimized, and the effectiveness of subsequent model training is improved.

[0011] Furthermore, the model prediction module is based on Python to build a machine learning regression model, conducts multi-model comparative training, and completes model iterative optimization through feature selection and hyperparameter optimization to achieve accurate prediction and real-time inference of humidity data, ensuring the model's prediction accuracy and inference speed.

[0012] Furthermore, the intelligent terminal interaction module independently completes front-end and back-end development, uses Flask to build back-end services, develops a visual interactive interface, realizes real-time access to data from the biomimetic flexible sensing module, online prediction of humidity data, and dynamic display of monitoring results, thereby improving the interactivity and ease of use of the system.

[0013] Furthermore, the system constructs a full-process control system, from biomimetic flexible sensor signal acquisition, image algorithm processing, feature extraction, to machine learning model training and optimization, and then to front-end and back-end system development and engineering implementation, to achieve seamless connection and integrated management of each link.

[0014] The beneficial effects of this invention are: 1. The present invention discloses an intelligent real-time humidity monitoring system based on biomimetic flexible sensing. It establishes a dedicated humidity data acquisition and algorithm analysis system to address the nonlinear signal characteristics of the biomimetic flexible double-layer sensor. This solves the problem of difficult signal processing of flexible sensors in traditional systems, fully leverages the advantages of curved surface fitting and high sensitivity of the biomimetic flexible sensor, and improves the sensor's scene adaptability. 2. The intelligent real-time humidity monitoring system based on biomimetic flexible sensing in this invention combines a self-designed curvature recognition algorithm with traditional digital image algorithms to complete grayscale normalization, adaptive threshold segmentation, and morphological filtering of the sensor response visualization image. It innovatively extracts multi-dimensional features such as surface curvature and texture distribution, effectively mining the core features related to humidity. The dataset of 800+ multi-gradient humidity samples provides high-quality data support for model training. 3. The present invention provides an intelligent real-time humidity monitoring system based on biomimetic flexible sensing. It uses a machine learning regression model built with Python. Through feature selection, multi-model comparison training and hyperparameter optimization, it achieves accurate prediction and real-time inference of humidity data. The model prediction average absolute error is as low as 1.8% RH, and the single-frame inference time is <35ms, which greatly improves the accuracy and real-time performance of humidity monitoring. 4. The present invention discloses an intelligent real-time humidity monitoring system based on biomimetic flexible sensing. It independently completes the front-end and back-end development of the real-time prediction system, uses Flask to build the back-end service, and develops a visual interactive interface. It realizes real-time access to sensor data, online humidity prediction and dynamic display of results, improves the interactivity and visualization of the system, and makes it convenient for users to grasp humidity monitoring data in real time. Attached Figure Description

[0015] Figure 1 Overall design structure diagram of an intelligent real-time humidity monitoring system based on biomimetic flexible sensing.

[0016] Figure 2 Design structure block diagram of the algorithm processing module of the intelligent humidity real-time monitoring system based on biomimetic flexible sensing.

[0017] Figure 3 Workflow diagram of the model prediction module of an intelligent humidity real-time monitoring system based on biomimetic flexible sensing.

[0018] To enable those skilled in the art to better understand the technical solutions of the present invention, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings. Other similar embodiments made by those skilled in the art based on the embodiments in this application without creative effort should all fall within the scope of protection of this application. Currently, accurate real-time monitoring of environmental humidity has become an important requirement in the field of intelligent sensing. Bionic flexible sensors have become a research hotspot due to their flexible fit advantages, but their nonlinear signal characteristics and the technical shortcomings of existing monitoring systems limit their practical application. Therefore, we designed an intelligent real-time humidity monitoring system based on bionic flexible sensing, built a full-process data processing and analysis system, combined with machine learning to achieve accurate real-time humidity prediction, completed the front-end and back-end system development and engineering implementation, effectively solved the problems of difficult signal processing, low monitoring accuracy, and poor real-time performance of flexible sensors, and improved the technical level and engineering application capabilities of intelligent humidity monitoring.

[0019] like Figure 1 As shown, the entire system consists of a biomimetic flexible sensing module, a data acquisition module, an algorithm processing module, a model prediction module, and a smart terminal interaction module; and as... Figure 2 As shown, the algorithm processing module covers three core processes: image preprocessing, feature extraction, and noise filtering; as Figure 3 As shown, the model prediction module includes four workflows: dataset construction, model training, hyperparameter optimization, and real-time inference.

[0020] As shown in Figure 3, the biomimetic flexible sensing module adopts a biomimetic flexible double-layer sensor structure. It utilizes the humidity response characteristics of flexible materials to achieve high sensitivity sensing of different humidity environments. At the same time, thanks to the flexible curved surface fitting characteristics, it can be adapted to monitoring scenarios of different shapes such as industrial pipelines, agricultural greenhouse curved surfaces, and home cabinets. The sensor outputs a visualized image signal of humidity response, providing raw data for subsequent data processing and analysis.

[0021] like Figure 3 As shown, the data acquisition module establishes a high-speed data transmission channel to acquire the visualized image signals output by the bionic flexible sensor in real time. The raw image data is transmitted to the algorithm processing module without delay or loss, while the initial classification and storage of the data are achieved. This ensures the integrity, real-time performance, and orderliness of the data acquisition, laying the foundation for subsequent algorithm processing.

[0022] like Figure 2As shown, the algorithm processing module first performs image preprocessing, using traditional digital image algorithms to sequentially perform grayscale normalization on the sensor response visualization image to eliminate image brightness differences; then it performs adaptive threshold segmentation to achieve accurate separation between the sensor response area and the background area; finally, it performs morphological filtering to effectively eliminate salt-and-pepper noise and Gaussian noise in the image, significantly improving the quality of the sensor response image.

[0023] like Figure 2 As shown, the algorithm processing module independently designed a curvature recognition algorithm. Combining high-quality image data after image preprocessing, it innovatively extracts multi-dimensional feature parameters such as surface curvature and texture distribution of the image based on the curved surface characteristics of the biomimetic flexible sensor. Through feature correlation analysis, it mines core features that are strongly correlated with humidity changes, eliminates invalid features, and provides high-quality feature input for model training.

[0024] like Figure 2 As shown, the algorithm processing module conducts experimental data collection under different humidity gradient environments based on the extracted multi-dimensional core features, constructing a multi-gradient humidity sample dataset of 800+ groups. Through data cleaning, precise noise filtering, and feature standardization, the quality of the dataset is optimized to ensure its diversity and effectiveness, providing solid data support for machine learning model training.

[0025] like Figure 2 As shown, the model prediction module is based on the Python development environment. It builds various machine learning regression models and conducts comparative training. The input feature set is further optimized through feature selection algorithm. Hyperparameter optimization is carried out by methods such as grid search and random search. After multiple rounds of iterative optimization of the model, the optimal model architecture is finally determined, which realizes accurate prediction and real-time inference of humidity data. The model prediction average absolute error is as low as 1.8% RH, and the single-frame inference time is <35ms, which meets the performance requirements of real-time monitoring.

[0026] like Figure 2 As shown, the intelligent terminal interaction module independently completes the integrated front-end and back-end development. The back-end uses the Flask framework to build a lightweight service, achieving seamless integration with the data acquisition module and model prediction module, supporting real-time access to biomimetic flexible sensor data and rapid transmission of humidity prediction results. The front-end develops a visual interactive interface, designing functions such as real-time display of humidity data, historical trend curves, and abnormal warning prompts, realizing dynamic visualization of online humidity prediction and monitoring results, and improving the user experience.

[0027] like Figure 2As shown, the entire system achieves fully autonomous control. From the design of curvature recognition algorithms for the characteristics of biomimetic flexible sensors, the fine preprocessing of sensor response images, multi-dimensional feature extraction and dataset construction, to the comparative training, hyperparameter optimization and iterative upgrade of machine learning models, and then to the construction of backend services based on Flask, the development of front-end visual interactive interfaces and the overall engineering implementation of the system, each link is seamlessly connected to form an integrated intelligent real-time humidity monitoring system. The system can be flexibly deployed in multiple scenarios such as industry, agriculture, and home, and has good stability and portability.

[0028] All features disclosed in this specification, or steps in all disclosed methods or processes, may be combined in any way, except for mutually exclusive features and / or steps. Any feature disclosed in this specification (including any appended claims and abstract) may be replaced by other equivalent or similar features unless specifically stated otherwise. That is, unless specifically stated otherwise, each feature is merely one example of a series of equivalent or similar features.

[0029] The above technical solution is only one embodiment of the present technical solution. Any changes that may be made by those skilled in the art to certain parts of it reflect the principle of the design system and should also fall within the scope of protection of this patent.

Claims

1. An intelligent real-time humidity monitoring system based on biomimetic flexible sensing, characterized in that, The real-time humidity monitoring system employs biomimetic flexible sensing and machine learning algorithms to achieve high-precision, low-latency real-time monitoring of environmental humidity. It includes a biomimetic flexible sensing module, a data acquisition module, an algorithm processing module, a model prediction module, and a smart terminal interaction module. The biomimetic flexible sensing module comprises a biomimetic flexible dual-layer sensor and a humidity response visualization image output unit. The algorithm processing module includes an image preprocessing unit, a curvature recognition unit, a feature extraction unit, a noise filtering unit, and a dataset construction unit. The model prediction module includes a Python machine learning regression model, a hyperparameter optimization unit, and a real-time inference unit. The smart terminal interaction module includes a Flask backend service unit, a visualization interface unit, and a real-time data display unit. The data acquisition module receives image signals from the biomimetic flexible sensing module and transmits them to the algorithm processing module. The algorithm processing module performs grayscale normalization, adaptive threshold segmentation, and morphological filtering on the visualized image, and extracts multi-dimensional features of surface curvature and texture distribution to construct 800+ The system consists of a multi-gradient humidity sample dataset; the model prediction module inputs the processed feature data into the trained regression model to achieve accurate humidity prediction and rapid inference; and the intelligent terminal interaction module receives the humidity prediction results and performs dynamic data display and online monitoring.

2. The intelligent humidity real-time monitoring system based on biomimetic flexible sensing as described in claim 1, characterized in that, The algorithm processing module is connected to the data acquisition module, receives sensor visualization image data, and performs image preprocessing, feature extraction, and noise filtering.

3. The intelligent humidity real-time monitoring system based on biomimetic flexible sensing as described in claim 1, characterized in that, The model prediction module is connected to the algorithm processing module, which receives multi-dimensional feature data to perform humidity calculation and real-time inference. The model prediction average absolute error is as low as 1.8% RH, and the single-frame inference time is <35ms, realizing real-time and accurate monitoring of environmental humidity.

Images