Intelligent monitoring nursing bed cleaning control method and system, electronic equipment and storage medium

Abstract

The invention relates to an intelligent monitoring nursing bed cleaning control method and system, electronic equipment and a storage medium. The method comprises the following steps: acquiring operation information of a nursing bed and work and rest information of a target object in real time; wherein the operation information comprises humidity information and pressure information; under the condition that the humidity information is greater than a humidity threshold value, determining a high-humidity area according to the humidity information; performing infrared scanning on the high-humidity area to obtain a reflection spectrum under a specific wavelength; based on the reflection spectrum, combining a convolutional neural network and a spectrum database to obtain a substance classification result; and according to a substance classification result, triggering a cleaning and drying process and dynamically adjusting cleaning parameters to complete cleaning of the nursing bed. Thus, the pollutants can be accurately classified, the pollutants are distinguished from common water stains, sweat and other interferents, and the problems that single humidity detection is prone to misjudgment and false triggering of cleaning are solved; hydroelectric resource waste and ineffective loss of equipment caused by cleaning with fixed cleaning parameters are reduced, and meanwhile, the cleaning and drying effects are guaranteed.

Description

Technical Field

[0001] This application relates to the field of nuclear safety technology, and in particular to an intelligent monitoring and control method, system, electronic device and storage medium for nursing bed cleaning. Background Technology

[0002] With the accelerating aging of the population, the demand for home and institutional care for disabled and semi-disabled patients is becoming increasingly prominent. As a core device for improving nursing efficiency and ensuring patients' quality of life, the functionality and intelligence of intelligent nursing beds have attracted much attention. Among these, the cleaning and drying function, as one of the key functions of a nursing bed, is directly related to patient hygiene and comfort. However, existing control methods still have significant limitations, making it difficult to balance timeliness, accuracy, and equipment durability. Specifically, this manifests in the inherent defects of the following two mainstream control modes: One type is the manual trigger control mode, which relies entirely on nurses to manually activate the cleaning and drying function after discovering patient contaminants through visual observation and regular inspections. However, the response is slow, and cleaning is easily delayed due to limited nurses' energy and untimely inspections. This not only significantly increases the workload of nurses but may also lead to nursing risks such as skin infections and bed odors caused by prolonged contaminant residue, seriously affecting nursing safety and patient comfort. The other type is the fixed-time trigger control mode, which automatically starts the cleaning process by preset fixed cleaning cycles (such as cleaning every 4 hours). This may not be able to respond to cleaning needs in a timely manner, resulting in contaminant residue and affecting nursing hygiene. Summary of the Invention

[0004] Therefore, it is necessary to provide an intelligent monitoring and control method, system, electronic device, and storage medium for the cleaning of nursing beds to address the aforementioned technical problems.

[0005] In a first aspect, this application provides an intelligent monitoring and control method for the cleaning of a nursing bed, the method comprising: The system acquires real-time operational information of the nursing bed and the daily routine information of the target patient; wherein the operational information includes humidity information and pressure information. If the humidity information is greater than the humidity threshold, a high humidity area is determined based on the humidity information; Infrared scanning was performed on the high humidity area to obtain the reflectance spectrum at a specific wavelength; Based on the reflected spectrum, combined with a convolutional neural network and a spectral database, the substance classification results are obtained; Based on the material classification results, a cleaning and drying process is triggered and cleaning parameters are dynamically adjusted to complete the cleaning of the nursing bed.

[0006] In one embodiment, the method further includes: When the humidity information is less than or equal to the humidity threshold, a cleaning prediction result is obtained based on the rest information and the pressure information, combined with the bed rest behavior analysis model; If the cleaning prediction result meets the cleaning threshold, determine whether the life protection condition is met. If the life protection conditions are met, the cleaning and drying process is triggered to complete the cleaning of the nursing bed.

[0007] In one embodiment, determining the high humidity area based on the humidity information when the humidity information is greater than a humidity threshold includes: If the humidity information is greater than the humidity threshold, a humidity thermal distribution map of the detection area is generated; The high humidity region is selected from the humidity thermal distribution map and its characteristic information is obtained; wherein, the characteristic information includes location information, area information and humidity peak value.

[0008] In one embodiment, obtaining the substance classification result based on the reflectance spectrum, combined with a convolutional neural network and a spectral database, includes: The target spectral vector is obtained by performing baseline correction, noise removal, and normalization on the spectral vector corresponding to the reflectance spectrum. The target spectral vector is input into a lightweight CNN to obtain the target probability distribution of multiple substances; wherein, the substances include feces, urine, ordinary water stains, sweat and / or cleaning residue; The substance category corresponding to the maximum probability in the target probability distribution is determined as the first classification result; Calculate the cosine similarity between the target spectral vector and the standard spectral vector in the spectral database, and obtain the second classification result based on the cosine similarity; If the first classification result and the second classification result are consistent, the first classification result is determined to be the substance classification result.

[0009] In one embodiment, the training method of the CNN includes: Obtain standard samples with labeled substance categories; wherein, the standard samples include fecal samples, urine samples and / or interference samples; The training set from the standard samples is input into the initial network model to obtain the predicted probability distribution; Based on the cross-entropy loss between the predicted and actual values ​​in the predicted probability distribution, backpropagation is used to update the network parameters iteratively. After each iteration, the model accuracy is evaluated based on the validation set in the standard samples until the model accuracy is greater than a reference threshold, thus obtaining the CNN.

[0010] In one embodiment, obtaining the cleaning prediction result based on the rest information and the stress information, combined with a bedridden behavior analysis model, includes: Based on the rest information and the stress information, core rest indicators are determined; wherein, the core rest indicators include single bed rest duration, daily cumulative bed rest duration and / or bed rest frequency; The core daily routine indicators are input into the bedridden behavior analysis model, and the cleanliness probability is output. The cleaning prediction result is obtained based on the cleaning probability.

[0011] In one embodiment, the lifetime protection condition is that the number of cleaning operations within a predetermined time window is less than the number of protection operations and the cleaning interval is longer than the protection duration.

[0012] Secondly, this application also provides an intelligent monitoring and control system for nursing bed cleaning, comprising: The first acquisition module is used to acquire the operating information of the nursing bed and the rest information of the target object in real time; wherein, the operating information includes humidity information and pressure information; The determination module is used to determine a high humidity area based on the humidity information when the humidity information is greater than a humidity threshold. The second acquisition module is used to perform infrared scanning on the high humidity area to obtain the reflection spectrum at a specific wavelength; The analysis module is used to obtain the substance classification results based on the reflectance spectrum, combined with a convolutional neural network and a spectral database; The cleaning control module is used to trigger the cleaning and drying process and dynamically adjust the cleaning parameters based on the material classification results to complete the cleaning of the nursing bed.

[0013] Thirdly, this application also provides an electronic device, including a memory and a processor, wherein the memory stores a computer program, and when the computer program is executed by the processor, it implements the steps of the method described in any embodiment of this application.

[0014] Fourthly, this application also provides a computer-readable storage medium having a computer program stored thereon, wherein the computer program, when executed by a processor, implements the steps of the method described in any embodiment of this application.

[0015] In the aforementioned intelligent monitoring and nursing bed cleaning control method, humidity thresholds are used as trigger nodes to achieve precise and automated closed-loop control of contaminant detection and cleaning. On the one hand, by first locating high-humidity areas and then performing targeted infrared scanning, combined with convolutional neural networks and spectral databases, contaminants are accurately classified, distinguishing them from ordinary water stains, sweat, and other interfering substances. This improves the problem of misjudgment and false triggering of cleaning by single humidity detection, significantly enhancing detection accuracy. On the other hand, cleaning parameters can be dynamically adjusted based on accurate substance classification results to achieve targeted cleaning, reducing the waste of water and electricity resources and ineffective equipment wear caused by fixed cleaning parameters. At the same time, it ensures cleaning and drying effects, reducing skin infections and odor generation caused by contaminant residues from the source, and improving the comfort and hygiene safety of bedridden patients. In addition, the entire detection, classification, and cleaning process is automatically triggered and executed, reducing manual intervention and cleaning operations, significantly reducing the workload of nursing staff. It is suitable for unattended or lightly attended nursing scenarios for disabled, semi-disabled, and other long-term bedridden individuals, making the cleaning function of the nursing bed more in line with actual nursing needs, and taking into account intelligence, practicality, and nursing professionalism. Attached Figure Description

[0016] Figure 1 This is a flowchart illustrating an intelligent monitoring and nursing bed cleaning control method according to an exemplary embodiment; Figure 2 This is a structural block diagram of an intelligent monitoring and nursing bed cleaning control system according to an exemplary embodiment; Figure 3 This is an internal structural diagram of an electronic device according to an exemplary embodiment. Detailed Implementation

[0017] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.

[0018] The terms "first," "second," and "third" used in the embodiments of this application are for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of indicated technical features. Therefore, a feature defined as "first," "second," or "third" may explicitly or implicitly include at least one of that feature. In the description of this application, "multiple" means at least two, such as two, three, etc., unless otherwise explicitly specified.

[0019] The intelligent monitoring and nursing bed cleaning control method provided in this application embodiment can be applied to electronic devices. The electronic devices can communicate with the intelligent monitoring and nursing bed wirelessly or via wired connection, or be built into the intelligent monitoring and nursing bed. The electronic devices can be any mobile terminal or fixed terminal. The terminal can be a device that provides voice and / or data connectivity to the user. For example, the terminal can be an Internet of Things (IoT) terminal, such as a sensor device, a mobile phone or so-called "cellular" phone, and a computer with an IoT terminal; for example, it can be a fixed, portable, pocket-sized, handheld, or computer-embedded device. In related technologies, electronic devices have interactive components, such as a touch screen or a non-touch screen.

[0020] In some embodiments, such as Figure 1 As shown, an intelligent monitoring and control method for nursing bed cleaning is provided, the method comprising the following steps: S101, real-time acquisition of the operation information of the nursing bed and the rest information of the target object; wherein, the operation information includes humidity information and pressure information.

[0021] In this embodiment of the application, the operation information indicates that during the operation of the nursing bed, monitoring data related to the bed status and the contact status between the target object and the bed are collected in real time by various sensors.

[0022] In this embodiment, the humidity information indicates humidity-related data of the nursing bed mattress surface and contact area collected by a humidity sensor / humidity sensor array, including dimensions such as single-point humidity value, humidity distribution, and humidity change rate.

[0023] In this embodiment, the pressure information indicates the relevant data of the pressure on the nursing bed body collected by the pressure sensor / pressure sensor array, including the pressure value of each measuring point, the overall average pressure value of the bed body, pressure distribution, pressure change trend and other dimensions, which are used to help determine the bed rest status of the target object.

[0024] In this embodiment, the rest information is based on data such as pressure information from the nursing bed and motion sensor data, which are statistically obtained to reflect the target subject's bed rest patterns. Examples include the cumulative bed rest time and the number of times the target subject lies in bed that day.

[0025] S102, if the humidity information is greater than the humidity threshold, determine the high humidity area based on the humidity information.

[0026] In this embodiment, the high humidity area indicates the core detection area on the surface of the nursing bed mattress where contaminants are suspected to exist.

[0027] In some embodiments, determining a high humidity region based on the humidity information when the humidity information is greater than a humidity threshold includes: If the humidity information is greater than the humidity threshold, a humidity thermal distribution map of the detection area is generated; The high humidity region is selected from the humidity thermal distribution map and its characteristic information is obtained; wherein, the characteristic information includes location information, area information and humidity peak value.

[0028] In this embodiment, the humidity thermal distribution map is based on the humidity sensor array grid of the nursing bed mattress. The real-time humidity value of each grid measuring point is mapped into a data-driven two-dimensional matrix or a visual color spectrum distribution carrier. The humidity difference in different areas of the mattress is intuitively represented by numerical / color gradients, which makes it easy to locate high humidity areas.

[0029] In this embodiment, the feature information indicates core quantitative data used to accurately describe the physical properties and humidity state of high-humidity areas. The feature information may include, but is not limited to, location information, area information, and / or humidity peaks.

[0030] In this embodiment of the application, the area information represents the quantitative data of the actual coverage of the high humidity area, which is obtained by converting the number of continuous grids in the high humidity area with the area of ​​a single grid.

[0031] In this embodiment, the humidity peak value indicates the maximum real-time humidity value among all grid measuring points in the high humidity area, reflecting the degree of liquid penetration in the area. It is a reference parameter to help determine the level of pollutants and adjust the cleaning water pressure.

[0032] In some embodiments, the electronic device can generate a unified grid coordinate system for the mattress based on humidity information. For example, with the upper left corner of the headboard of the nursing bed as the origin (0,0), a two-dimensional grid rectangular coordinate system for the mattress is established: the horizontal axis (headboard → footboard) is the X-axis, and the vertical axis (left side of the bed → right side of the bed) is the Y-axis. Each 5cm × 5cm grid is a coordinate node. For example, the grid coordinates of the 3rd grid on the X-axis and the 5th grid on the Y-axis are (3,5), thus achieving unique coordinate calibration of all humidity measurement points on the mattress. Using the mattress grid coordinates as rows / columns, the pre-processed humidity values ​​of each grid are filled into the corresponding positions in the matrix to form a humidity value matrix that corresponds one-to-one with the mattress grid. Through gradient color mapping rules (such as blue = ≤50%RH, yellow = 50%-70%RH, red => 70%RH), the humidity values ​​of the data matrix are converted into a color spectrum, which can be displayed in real time on the nursing bed touch panel and mobile terminal APP, making it convenient for caregivers to intuitively view the location of contaminated areas.

[0033] In some embodiments, a preset humidity threshold (e.g., 70%RH) is used as the criterion. The data of the humidity thermal distribution map is traversed in a two-dimensional matrix to filter out all grids whose humidity information is greater than the humidity threshold. These grids are marked as candidate high humidity grids, and their coordinates and humidity values ​​are recorded. The candidate high humidity grids are clustered using a neighborhood connectivity analysis algorithm. For example, it is determined whether the eight adjacent grids (upper, lower, left, right, upper left, upper right, lower left, and lower right) of each candidate grid are candidate high humidity grids. If they are a continuous and connected cluster of candidate grids, they are determined to be a valid high humidity region. If they are isolated single-point candidate grids (without any adjacent candidate grids), they are removed and not included in the high humidity region.

[0034] In this embodiment, a thermal distribution map is used to visually present the humidity distribution of the detection area. Combined with screening rules, scattered humidity interference points exceeding the threshold (such as water droplets and sweat) are eliminated, reducing subsequent invalid detections caused by non-polluting humidity anomalies and accurately locating the core high-humidity area of ​​suspected contaminants. Furthermore, by extracting quantitative feature information such as location, area, and humidity peak, clear directional positioning and scanning range limits are provided for subsequent infrared scanning, avoiding the problems of computational redundancy and low detection efficiency caused by full-area infrared scanning. At the same time, core quantitative data support is provided for the subsequent dynamic adjustment of cleaning parameters based on the actual location, coverage, and liquid penetration degree of the contaminated area, making subsequent spectral detection and cleaning and drying actions more targeted, significantly reducing the ineffective consumption of water and electricity resources and equipment computing power, and taking into account the accuracy of nursing bed contamination detection, the efficiency of subsequent execution links, and the energy saving of equipment operation, thereby improving the intelligence and adaptability of nursing bed cleaning control from the source.

[0035] S103, perform infrared scanning on the high humidity area to obtain the reflection spectrum at a specific wavelength.

[0036] In this embodiment, the infrared scanning indicator emits infrared light of a specific wavelength range in a designated area (a high-humidity area in this scheme) in a directional manner, while simultaneously receiving the infrared light reflected from the area under test and collecting the reflected light spectral data.

[0037] In this embodiment, the reflectance spectrum indicates the set of values ​​and corresponding curves of infrared light reflectance at different specific wavelengths after infrared light is irradiated onto the measured area (high humidity area). Different substances will form reflectance spectra with unique characteristics due to their different molecular structures, which is the core basis for substance classification.

[0038] S104. Based on the reflected spectrum, combined with a convolutional neural network and a spectral database, the substance classification result is obtained.

[0039] In this embodiment of the application, the spectral database refers to a pre-constructed database that stores standard reflectance spectral data of various typical substances in the nursing bed scenario, including the standard normalized reflectance spectrum of the substance, characteristic peak information and corresponding category labels.

[0040] For example, the spectral database can be a database that stores standard reflectance spectra of feces, urine, ordinary water stains, sweat, and cleaning fluid residue, and labels each type of spectrum with corresponding category labels.

[0041] In some embodiments, obtaining the substance classification result based on the reflectance spectrum, combined with a convolutional neural network and a spectral database, includes: The target spectral vector is obtained by performing baseline correction, noise removal, and normalization on the spectral vector corresponding to the reflectance spectrum. The target spectral vector is input into a lightweight CNN to obtain the target probability distribution of multiple substances; wherein, the substances include feces, urine, ordinary water stains, sweat and / or cleaning residue; The substance category corresponding to the maximum probability in the target probability distribution is determined as the first classification result; Calculate the cosine similarity between the target spectral vector and the standard spectral vector in the spectral database, and obtain the second classification result based on the cosine similarity; If the first classification result and the second classification result are consistent, the first classification result is determined to be the substance classification result.

[0042] In some embodiments, after infrared scanning of a high-humidity area, a set of reflectance values ​​at each wavelength is obtained, resulting in a spectral vector. The spectral vector is then fitted and subtracted using a second-order polynomial baseline fitting method to obtain a first spectral vector. A Savitzky-Golay (SG) smoothing filter (e.g., window size 11, polynomial order 2) is used to filter out random noise collected by the sensor, resulting in a second spectral vector. Alternatively, the second spectral vector can be standardized using max-min normalization to obtain a target spectral vector.

[0043] In some embodiments, the adapted target vector is input into the quantized lightweight CNN, and then passes through the forward computation of convolution, activation, pooling, and fully connected layers in sequence. Finally, the output layer result is mapped to the probability value in the interval [0,1] by the Softmax activation function, and the sum of all probability values ​​is 1. The most likely material category is extracted from the target probability distribution, that is, the material category corresponding to the maximum probability, as the first classification result of the CNN model.

[0044] In some embodiments, the training method of the CNN includes: Obtain standard samples with labeled substance categories; wherein, the standard samples include fecal samples, urine samples and / or interference samples; The training set from the standard samples is input into the initial network model to obtain the predicted probability distribution; Based on the cross-entropy loss between the predicted and actual values ​​in the predicted probability distribution, backpropagation is used to update the network parameters iteratively. After each iteration, the model accuracy is evaluated based on the validation set in the standard samples until the model accuracy is greater than a reference threshold, thus obtaining the CNN.

[0045] In this embodiment of the application, the interfering sample may include, but is not limited to, sweat sample, detergent residue sample and / or ordinary water stain sample.

[0046] In some embodiments, multiple types of standard samples are collected according to the actual application scenario of the nursing bed, including fecal samples, urine samples, and interference samples. The interference samples include sweat samples, detergent residue samples, and ordinary water stain samples. Multiple fecal and urine samples are collected, such as 100 samples, and more than 50 samples of each type of interference sample are collected. All samples simulate the actual use scenario of bedridden patients. For example, fecal samples cover different types of consistency and water content, urine samples cover different types of concentration and pH, sweat samples simulate the sweat composition of patients when they are bedridden, detergent residue samples are prepared using cleaning solution provided with the nursing bed, and ordinary water stain samples are drinking water. All samples are collected on the same mattress material as the nursing bed to ensure background consistency.

[0047] In some embodiments, all standard samples that have been labeled and preprocessed are hierarchically divided into training and validation sets in a 7:3 ratio. This hierarchical division avoids sample class imbalance and ensures that the sample proportion of each material category in the two datasets is consistent. The initial network model adopts a lightweight one-dimensional convolutional neural network customized for spectral detection of nursing beds. The network input dimension matches the dimension of the preprocessed spectral data, and the output layer is set with neurons corresponding to the number of material categories. First, the spectral data in the training set is adjusted to conform to the input specification of the initial network model. Then, the training set samples are input into the initial network model in batches according to the set batch size. The samples are processed by the model's convolutional layer, activation layer, pooling layer, and fully connected layer for feature extraction and fusion. Finally, the features are mapped to values ​​in the [0,1] interval through the Softmax activation function of the output layer, forming the multi-category predicted probability distribution of each batch of samples. The sum of the probability values ​​of all categories is 1.

[0048] In some embodiments, for multi-substance classification tasks, a multi-class cross-entropy loss function is used to calculate the loss value between the predicted value in the predicted probability distribution of the training set samples and the true value of the sample labels in batches. The magnitude of the loss value reflects the degree of deviation between the current prediction result of the model and the actual situation. The Adam optimizer is selected and a learning rate adapted to the lightweight model is set. The model is iterated in a mini-batch training manner. Each batch of training set samples is input, the loss value is calculated once, and then the loss value is backpropagated from the model output layer to the input layer. The gradient of the parameters of each network layer is calculated. According to the gradient descent principle, the weights and bias parameters of all network layers such as convolutional layers and fully connected layers are updated according to the preset rules of the optimizer to complete one round of network parameter update. After all batches of training set samples have been input, loss calculation and parameter update are completed, one round of model iteration is completed.

[0049] In some embodiments, classification accuracy is used as the core evaluation metric for model precision. After each iteration, all standard samples in the validation set are input into the updated network model to obtain the predicted material categories of the validation set samples. The predicted results are compared with the true labels of the samples one by one, and the proportion of correctly predicted samples to the total number of samples in the validation set is counted as the precision value of the current model. A preset model precision reference threshold is set. If the model precision after a single iteration does not reach the threshold, the next iteration training continues. If the model precision consistently reaches or exceeds the reference threshold for multiple iterations, the model training is considered complete, the iteration stops, and the final network parameters of the model at this point are saved to obtain the basic CNN model. At the same time, the model is subjected to INT8 quantization to simplify the model's computation and reduce its size, forming a lightweight CNN model that can be directly deployed in the embedded main control system of the nursing bed.

[0050] In some embodiments, based on the first classification result obtained in the previous stage, all standard spectral vectors corresponding to the category are retrieved from the spectral database. Cosine similarity is calculated between the target spectral vector and these standard spectral vectors one by one. Then, the mean of all calculation results is calculated and compared with a preset similarity threshold. If the mean reaches or exceeds the threshold, the second classification result is determined to be in the same category as the first classification result. If the mean does not reach the threshold, the second classification result is determined to be an unknown substance.

[0051] In some embodiments, the consistency of the first classification result and the second classification result is checked. If the corresponding substance categories are exactly the same, the substance category is determined as the final substance classification result. If the corresponding substance categories are different, the substance is directly marked as an unknown pollutant and is not considered a valid substance classification result.

[0052] In this embodiment, a dual classification verification mechanism combining CNN model inference and cosine similarity verification of spectral database is constructed to reduce the identification misjudgment problem that is prone to occur in single algorithm models, and significantly improve the classification accuracy of feces, urine and ordinary water stains, sweat and other interfering liquids. Furthermore, by retrieving the corresponding category standard spectral vector based on the first classification result to calculate similarity, the computational redundancy caused by comparing with the standard vector of all categories is reduced, perfectly adapting to the computational power limitations of the embedded main control of the nursing bed. At the same time, the quantitative judgment of similarity threshold makes the second classification result more rigorous, and the consistency verification of the dual results further reduces the misjudgment rate of material classification to a minimum, reducing the inappropriate cleaning and drying process triggered by incorrect classification, and reducing the waste of water and electricity resources and the ineffective wear and tear of the cleaning module. Ultimately, it provides a highly reliable basis for determining the material category for subsequent accurate triggering of the cleaning and drying process and dynamic adaptation of cleaning parameters, so that the contaminant identification link of the nursing bed has both accuracy, reliability and equipment operating efficiency, and is highly adaptable to the actual nursing needs of disabled and semi-disabled patients.

[0053] S105, based on the material classification results, trigger the cleaning and drying process and dynamically adjust the cleaning parameters to complete the cleaning of the nursing bed.

[0054] In some embodiments, after obtaining effective and accurate material classification results, the nursing bed main control system will directly trigger the fully automatic cleaning and drying linkage process. Simultaneously, combining the material classification results with the previously extracted high-humidity area feature information, the system will dynamically adjust the core parameters of the entire cleaning and drying process. First, based on the location information of the high-humidity area, a command is sent to the nozzle drive module of the cleaning unit to control the swingable nozzle to precisely move to the center of the contaminated area. Then, based on the area information, the maximum swing angle and swing range of the nozzle are adjusted to ensure that the nozzle covers the entire high-humidity area. Finally, based on the material classification results… The system retrieves matching basic cleaning parameters from the rule base. For fecal contaminants, it activates a high-pressure rinsing mode, adjusting the water pump pressure to a medium-high value, maintaining the rinsing water temperature at the target temperature, and extending the rinsing time. Simultaneously, it controls the nozzles to oscillate at a high frequency to achieve all-round cleaning. For urine contaminants, it activates a regular rinsing mode, using medium water pressure and the same constant water temperature, matching the regular rinsing time to the contaminated area, and oscillating the nozzles at a regular frequency. For interfering samples such as ordinary water stains, sweat, and cleaning residue, it activates a low-pressure gentle rinsing mode, reducing the water pump pressure and shortening the rinsing time to avoid over-cleaning. After the cleaning process is completed, the wastewater recycling module simultaneously collects the wastewater, and the main control system then triggers the drying process. Based on the material classification results and the humidity peak of the high humidity area, the drying parameters are adjusted. For fecal and urine contaminants, due to the high humidity peak and deep liquid penetration, a medium-high temperature drying mode is adopted, adjusting the drying temperature to the corresponding temperature and the fan to high speed, and the drying time is extended accordingly as the humidity peak increases. For interfering samples, due to the low humidity peak and no penetration, a low-temperature gentle drying mode is adopted, lowering the drying temperature and fan speed, and shortening the drying time. During the drying process, the mattress humidity sensor will collect the humidity data of the contaminated area in real time. When the humidity value is detected to drop to the preset cleaning qualification threshold, the drying process is completed.

[0055] In some embodiments, the electronic device can synchronously retrieve lifespan protection-related data from the device's operating storage module, such as the cumulative number of cleaning and drying operations within 24 hours, the duration of the last cleaning interval, the cumulative operating time of the core cleaning components (water pump, fan, heating element), and the current real-time operating status of the components (temperature, current). This constructs a triple matching mechanism of "substance contamination level - lifespan protection data - dynamic parameter adjustment," balancing cleaning effectiveness and equipment lifespan. The specific execution logic is as follows: First, the substance classification results are divided into three levels according to the degree of contamination and cleaning difficulty: heavily contaminated (feces), moderately contaminated (urine), and lightly contaminated (ordinary water stains, sweat, cleaning residue). The upper limits of the core component operating parameters corresponding to different contamination levels are preset. Heavily contaminated components are allowed to operate at their rated upper limits for a short period, moderately contaminated components operate at normal load, and lightly contaminated components operate at low load. After receiving the substance classification results, the main control system first determines the contamination level, and then performs a secondary parameter adjustment based on the lifespan protection data. If the number of cleaning operations within 24 hours does not reach the preset upper limit, or the last cleaning interval exceeds the preset upper limit, the system will adjust the parameters accordingly. If the minimum interval or cumulative running time of a component does not reach the warning value, cleaning and drying will be performed according to the original parameters matched to the contamination level to ensure cleaning effect. If the number of cleanings in 24 hours is close to the preset limit, the last cleaning interval is short, or the real-time operating temperature / current of the component is close to the warning value, even for heavy contamination, the water pump pressure and fan speed will be appropriately reduced to shorten the high-load running time while ensuring cleaning effect. In the drying process, the drying temperature will be appropriately reduced and the drying time will be extended to achieve cleaning and drying in a low-load manner. For light contamination, regardless of the life protection data, low-load cleaning and drying parameters will always be used to avoid meaningless high-load operation that aggravates component wear. At the same time, during the cleaning and drying process, the main control system will monitor the operating status of the core components in real time. If an abnormal increase in component temperature or current is detected, the relevant parameters will be fine-tuned immediately, and the process will continue after the status returns to normal. This reduces excessive wear and tear on components caused by solely pursuing cleaning effect and reduces incomplete cleaning caused by one-sided protection of equipment, achieving a dynamic balance between cleaning effect guided by material classification results and the service life of the nursing bed.

[0056] In the aforementioned intelligent monitoring and nursing bed cleaning control method, humidity thresholds are used as trigger nodes to achieve precise and automated closed-loop control of contaminant detection and cleaning. On the one hand, by first locating high-humidity areas and then performing targeted infrared scanning, combined with convolutional neural networks and spectral databases, contaminants are accurately classified, distinguishing them from ordinary water stains, sweat, and other interfering substances. This improves the problem of misjudgment and false triggering of cleaning by single humidity detection, significantly enhancing detection accuracy. On the other hand, cleaning parameters can be dynamically adjusted based on accurate substance classification results to achieve targeted cleaning, reducing the waste of water and electricity resources and ineffective equipment wear caused by fixed cleaning parameters. At the same time, it ensures cleaning and drying effects, reducing skin infections and odor generation caused by contaminant residues from the source, and improving the comfort and hygiene safety of bedridden patients. In addition, the entire detection, classification, and cleaning process is automatically triggered and executed, reducing manual intervention and cleaning operations, significantly reducing the workload of nursing staff. It is suitable for unattended or lightly attended nursing scenarios for disabled, semi-disabled, and other long-term bedridden individuals, making the cleaning function of the nursing bed more in line with actual nursing needs, and taking into account intelligence, practicality, and nursing professionalism.

[0057] In some embodiments, the method further includes: When the humidity information is less than or equal to the humidity threshold, a cleaning prediction result is obtained based on the rest information and the pressure information, combined with the bed rest behavior analysis model; If the cleaning prediction result meets the cleaning threshold, determine whether the life protection condition is met. If the life protection conditions are met, the cleaning and drying process is triggered to complete the cleaning of the nursing bed.

[0058] In some embodiments, the lifetime protection condition is that the number of cleaning operations within a predetermined time window is less than the number of protection operations and the cleaning interval is longer than the protection duration.

[0059] In some embodiments, obtaining the cleaning prediction result based on the rest information and the stress information, combined with a bedridden behavior analysis model, includes: Based on the rest information and the stress information, core rest indicators are determined; wherein, the core rest indicators include single bed rest duration, daily cumulative bed rest duration and / or bed rest frequency; The core daily routine indicators are input into the bedridden behavior analysis model, and the cleanliness probability is output. The cleaning prediction result is obtained based on the cleaning probability.

[0060] In some embodiments, the real-time bed rest status of the target object is first determined by pressure information, distinguishing between effective bed rest, temporary ambulation, and long-term ambulation, and accurately defining the start and end times of each effective bed rest. Then, combined with stored daily routine information, standardized core daily routine indicators are calculated based on the defined effective bed rest periods. These indicators include single bed rest duration (the time difference of a single effective bed rest period), cumulative bed rest duration for the day (the sum of the durations of all effective bed rest periods for the day), and bed rest frequency (the number of effective bed rests for the day or the number of bed rests per unit time). The above-mentioned quantified core daily routine indicators are normalized to form feature vectors that conform to the model input specifications. These vectors are then input into a pre-trained bed rest behavior analysis model. The model mines the correlation between the core daily routine indicators and preventive cleaning needs, outputs the cleaning probability that cleaning and drying need to be initiated, and uses this cleaning probability as the core basis to obtain the corresponding cleaning prediction result, providing a quantitative judgment standard for whether to trigger the cleaning process in the future.

[0061] In some embodiments, the cleaning probability in the cleaning prediction result is first compared with a preset cleaning threshold. If the cleaning probability reaches or exceeds the threshold, a comprehensive verification process for life protection conditions is immediately triggered. All actual operating data related to life protection are retrieved from the device operation storage module of the nursing bed, including the cumulative number of cleaning and drying processes executed in the past 24 hours, the time interval from the end of the last cleaning and drying to the present, the real-time operating parameters and cumulative working time of cleaning core components such as water pumps, fans, and heating elements. These actual data are then checked against preset life protection parameter thresholds one by one, including the upper limit of cleaning times in 24 hours, the minimum time interval between two cleanings, and the upper limit of rated operating parameters of core components. If all actual operating data meet the preset life protection parameter threshold requirements and none exceed the limit, the life protection conditions are deemed to be met; otherwise, the conditions are deemed not to be met.

[0062] It should be understood that although the steps in the flowcharts of the embodiments described above are shown sequentially according to the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless explicitly stated herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least some steps in the flowcharts of the embodiments described above may include multiple steps or multiple stages. These steps or stages are not necessarily completed at the same time, but can be executed at different times. The execution order of these steps or stages is not necessarily sequential, but can be performed alternately or in turn with other steps or at least some of the steps or stages of other steps.

[0063] Based on the same inventive concept, this application also provides an intelligent monitoring and nursing bed cleaning control system for implementing the intelligent monitoring and nursing bed cleaning control method described above. The solution provided by this system is similar to the solution described in the above method; therefore, the specific limitations in one or more embodiments of the intelligent monitoring and nursing bed cleaning control system provided below can be found in the limitations of the intelligent monitoring and nursing bed cleaning control method described above, and will not be repeated here.

[0064] In one embodiment, such as Figure 2 As shown, an intelligent monitoring and control system for nursing bed cleaning is provided, the system comprising: The first acquisition module 10 is used to acquire the operation information of the nursing bed and the rest information of the target object in real time; wherein, the operation information includes humidity information and pressure information; The determining module 20 is used to determine a high humidity area based on the humidity information when the humidity information is greater than a humidity threshold. The second acquisition module 30 is used to perform infrared scanning on the high humidity area to obtain the reflection spectrum at a specific wavelength; Analysis module 40 is used to obtain material classification results based on the reflectance spectrum, combined with a convolutional neural network and a spectral database; The cleaning control module 50 is used to trigger the cleaning and drying process and dynamically adjust the cleaning parameters according to the material classification results to complete the cleaning of the nursing bed.

[0065] Each module in the aforementioned intelligent monitoring and nursing bed cleaning control system can be implemented entirely or partially through software, hardware, or a combination thereof. Each module can be embedded in the processor of the electronic device in hardware form or independent of the processor, or it can be stored in the memory of the electronic device in software form, so that the processor can call and execute the corresponding operations of each module.

[0066] In one embodiment, an electronic device is provided, the internal structure of which can be shown as follows: Figure 3As shown, the electronic device includes a processor, memory, communication interface, display unit, and input system connected via a method bus. The processor provides computing and control capabilities. The memory includes a non-volatile storage medium and internal memory. The non-volatile storage medium stores operating methods and computer programs. The internal memory provides an environment for the operation of the operating methods and computer programs stored in the non-volatile storage medium. The communication interface is used for wired or wireless communication with external terminals; wireless communication can be achieved through Wi-Fi, mobile cellular networks, NFC (Near Field Communication), or other technologies. When the computer program is executed by the processor, it implements an intelligent monitoring and control method for cleaning a nursing bed. The display screen can be an LCD screen or an e-ink display screen. The input system can be a touch layer covering the display screen, or buttons, a trackball, or a touchpad mounted on the device's casing.

[0067] Those skilled in the art will understand that Figure 3 The structure shown is merely a block diagram of a portion of the structure related to the present application and does not constitute a limitation on the electronic device to which the present application is applied. The specific electronic device may include more or fewer components than shown in the figure, or combine certain components, or have different component arrangements.

[0068] In one embodiment, a computer-readable storage medium is provided having a computer program stored thereon that, when executed by a processor, implements the steps in the above method embodiments.

[0069] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium. When executed, the computer program can include the processes of the embodiments of the above methods. Furthermore, any references to memory, storage, databases, or other media used in the embodiments provided in this application can include at least one of non-volatile and volatile memory.

[0070] The technical features of the above embodiments can be combined arbitrarily. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as the combination of these technical features does not contradict each other, it should be considered within the scope of this specification. The above embodiments only illustrate several implementation methods of this application, and their descriptions are relatively specific and detailed, but they should not be construed as limiting the scope of this application. It should be noted that for those skilled in the art, several modifications and improvements can be made without departing from the concept of this application, and these all fall within the protection scope of this application. Therefore, the protection scope of this application should be determined by the appended claims.

Claims

1. A method for intelligent monitoring and controlling the cleaning of a nursing bed, characterized in that, The method includes: The system acquires real-time operational information of the nursing bed and the daily routine information of the target patient; wherein the operational information includes humidity information and pressure information. If the humidity information is greater than the humidity threshold, a high humidity area is determined based on the humidity information; Infrared scanning was performed on the high humidity area to obtain the reflectance spectrum at a specific wavelength; Based on the reflected spectrum, combined with a convolutional neural network and a spectral database, the substance classification results are obtained; Based on the material classification results, a cleaning and drying process is triggered and cleaning parameters are dynamically adjusted to complete the cleaning of the nursing bed.

2. The method according to claim 1, characterized in that, The method further includes: When the humidity information is less than or equal to the humidity threshold, a cleaning prediction result is obtained based on the rest information and the pressure information, combined with the bed rest behavior analysis model; If the cleaning prediction result meets the cleaning threshold, determine whether the life protection condition is met. If the life protection conditions are met, the cleaning and drying process is triggered to complete the cleaning of the nursing bed.

3. The method according to claim 1, characterized in that, When the humidity information is greater than a humidity threshold, determining the high humidity area based on the humidity information includes: If the humidity information is greater than the humidity threshold, a humidity thermal distribution map of the detection area is generated; The high humidity region is selected from the humidity thermal distribution map and its characteristic information is obtained; wherein, the characteristic information includes location information, area information and humidity peak value.

4. The method according to claim 1, characterized in that, The substance classification results obtained based on the reflectance spectrum, combined with a convolutional neural network and a spectral database, include: The target spectral vector is obtained by performing baseline correction, noise removal, and normalization on the spectral vector corresponding to the reflectance spectrum. The target spectral vector is input into a lightweight CNN to obtain the target probability distribution of multiple substances; wherein, the substances include feces, urine, ordinary water stains, sweat and / or cleaning residue; The substance category corresponding to the maximum probability in the target probability distribution is determined as the first classification result; Calculate the cosine similarity between the target spectral vector and the standard spectral vector in the spectral database, and obtain the second classification result based on the cosine similarity; If the first classification result and the second classification result are consistent, the first classification result is determined to be the substance classification result.

5. The method according to claim 4, characterized in that, The training methods for the CNN include: Obtain standard samples with labeled substance categories; wherein, the standard samples include fecal samples, urine samples and / or interference samples; The training set from the standard samples is input into the initial network model to obtain the predicted probability distribution; Based on the cross-entropy loss between the predicted and actual values ​​in the predicted probability distribution, backpropagation is used to update the network parameters iteratively. After each iteration, the model accuracy is evaluated based on the validation set in the standard samples until the model accuracy is greater than a reference threshold, thus obtaining the CNN.

6. The method according to claim 2, characterized in that, The step of obtaining a cleaning prediction result based on the rest information and stress information, combined with a bedridden behavior analysis model, includes: Based on the rest information and the stress information, core rest indicators are determined; wherein, the core rest indicators include single bed rest duration, cumulative bed rest duration on a given day, and / or bed rest frequency; The core daily routine indicators are input into the bedridden behavior analysis model, and the cleanliness probability is output. The cleaning prediction result is obtained based on the cleaning probability.

7. The method according to claim 2, characterized in that, The lifespan protection conditions are that the number of cleaning operations within a predetermined time window is less than the number of protection operations and the cleaning interval is longer than the protection duration.

8. An intelligent monitoring and control system for nursing bed cleaning, characterized in that, The system includes: The first acquisition module is used to acquire the operating information of the nursing bed and the rest information of the target object in real time; wherein, the operating information includes humidity information and pressure information; The determination module is used to determine a high humidity area based on the humidity information when the humidity information is greater than a humidity threshold. The second acquisition module is used to perform infrared scanning on the high humidity area to obtain the reflection spectrum at a specific wavelength; The analysis module is used to obtain the substance classification results based on the reflectance spectrum, combined with a convolutional neural network and a spectral database; The cleaning control module is used to trigger the cleaning and drying process and dynamically adjust the cleaning parameters based on the material classification results to complete the cleaning of the nursing bed.

9. An electronic device, characterized in that, The method includes a memory and a processor, wherein the memory stores a computer program, which, when executed by the processor, implements the method according to any one of claims 1 to 7.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it performs the steps of the method according to any one of claims 1 to 7.

Images