A method, system, medium, and device for drop monitoring based on radio frequency signals

Abstract

The application discloses a kind of based on radio frequency signal's drop monitoring method, system, medium and equipment, signal data returned by IR-UWB radar is collected;The influence of changing environment on drop signal is reduced by subtracting the average change along the time window of slow time;The signals of each interval are FFT, and the cross-correlation coefficient of low-frequency part and high-frequency part is calculated, the interval with drop signal has greater coefficient change, so as to locate the signal interval where drop is located;Signal interval signal is taken out according to time direction, input neural network model to predict drop frequency;Bidirectional LSTM module is used to learn time information, and binary cross-entropy is used to train neural network;Drop rate correction method is used to infer drop rate, and drop event is restored.The application realizes contactless monitoring of drop;It is not related to privacy problem, avoids the limitation of existing drop detection scheme, realizes the high-precision recovery of drop event.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of intelligent medical treatment, and particularly relates to a drop liquid monitoring method and system based on a radio frequency signal, a medium and equipment. BACKGROUND

[0002] Intravenous therapy (I.V.) is widely used in hospital standard clinical treatment methods. Safe and accurate intravenous infusion, i.e. administering drugs at a specific speed according to the disease, dosage and patient condition, plays a crucial role in the treatment process. Improper drop speed can lead to pulmonary edema, blood stream infection, and even death. Therefore, monitoring intravenous infusion while observing the patient's response is crucial to avoid improper drug treatment.

[0003] However, the most advanced intravenous therapy drug monitoring still faces many problems. Specifically, most hospitals have difficulty assigning nurses to carefully monitor the intravenous infusion treatment of all patients due to labor constraints. Therefore, automatic intravenous therapy drop monitoring has attracted great interest from researchers in recent years. Among them, the visual-based method is easy to deploy, but requires a high-resolution camera to work in good lighting conditions, and this method can raise significant privacy concerns.

[0004] In contrast, the sensing method using wired / wireless sensors (such as photoelectric sensors, RFID tags, etc.) has an advantage in privacy protection. However, the sensor needs to be redeployed and recalibrated every time, thereby introducing unnecessary labor costs. Therefore, it is worth further exploring to achieve accurate, tag-free and privacy-protected drop rate estimation. SUMMARY

[0005] The technical problem to be solved by the present application is to provide a drop liquid monitoring method and system based on a radio frequency signal, a medium and equipment to solve the technical problem of non-bound drop liquid monitoring in a complex environment, which uses wireless radio frequency signals to achieve tag-free and accurate drop rate estimation without the need for one-to-one pre-deployment of drop liquid, making it possible for users to adapt to medication.

[0006] The application adopts the following technical solutions:

[0007] A drop liquid monitoring method based on a radio frequency signal, comprising the following steps:

[0008] S1, collecting signal data and forming a signal matrix, and subtracting the average value along the t direction of the signal matrix;

[0009] S2, performing fast Fourier transform on the signal matrix whose average value is subtracted along the t direction in step S1 to obtain a frequency spectrum signal, dividing the frequency spectrum signal into a high frequency part and a low frequency part, and calculating a cross-correlation coefficient curve of the frequency spectrum signals of the high frequency part and the low frequency part;

[0010] S3, divide the cross-correlation coefficient curve obtained in step S2 into a plurality of windows, determine a distance interval corresponding to a cross-correlation coefficient curve with a maximum sum of window standard deviations, correspond the distance interval to a drop signal distance, and take out signals in the distance interval in a time direction and record as y(t0, i);

[0011] S4, input the signal y(t0, i) obtained in step S3 into a neural network to predict signal fluctuation peak position information caused by drops;

[0012] S5, determine a drop rate according to the signal fluctuation peak position information obtained in step S4 to realize drop monitoring based on a radio frequency signal.

[0013] Specifically, in step S1, an IR-UWB radar is used to collect signal data.

[0014] Specifically, in step S4, the neural network includes three residual convolution blocks, one bidirectional long short-time neural network block, and one fully connected block.

[0015] Further, the loss function of the neural network is binary cross entropy.

[0016] Further, the loss function L1 is:

[0017]

[0018] wherein r is an actual value of the peak position, is a prediction result.

[0019] Specifically, in step S5, a drop rate correction method is used to check whether all drop intervals are within a set range to determine the drop rate.

[0020] Further, the steps of the drop rate correction method are as follows:

[0021] The interval time between two adjacent drop events is calculated, and the median of all drop interval times is calculated, then the interval time between each two drop events is compared with the median, when the interval time between two adjacent drop events is greater than 1.1 times of the median, it is determined as missing, and the drop event with the maximum probability in the interval between the two adjacent drops is added as a supplement; when the interval time between two adjacent drop events is less than 0.9 times of the median, it is determined as a false judgment, and one of the two adjacent drop events is removed, so that the interval of the remaining drop events is at the median.

[0022] In a second aspect, an embodiment of the present application provides a drop monitoring system based on a radio frequency signal, comprising:

[0023] A collection module collects signal data and forms a signal matrix, and subtracts an average value from the signal matrix along a t direction;

[0024] a transform module, performing fast Fourier transform on the signal matrix subtracted by the average value along the t direction to obtain a frequency spectrum signal, dividing the frequency spectrum signal into a high frequency part and a low frequency part, and calculating a cross-correlation coefficient curve of the high frequency part and the low frequency part;

[0025] a positioning module, dividing the cross-correlation coefficient curve obtained by the transform module into a plurality of windows, determining a distance interval corresponding to the cross-correlation coefficient curve with the maximum sum of window standard deviations, corresponding the distance interval to a drop signal distance, and taking out signals in the distance interval in the time direction and recording as y(t0, i);

[0026] a training module, inputting the signal y(t0, i) obtained by the positioning module into a neural network to predict signal fluctuation peak position information caused by drops;

[0027] a correction module, determining a drop rate according to the signal fluctuation peak position information obtained by the training module to realize drop monitoring based on a radio frequency signal.

[0028] In a third aspect, a computer device includes a memory, a processor, and a computer program stored in the memory and executable on the processor, and the processor implements the steps of the drop monitoring method based on a radio frequency signal.

[0029] In a fourth aspect, an embodiment of the present application provides a computer readable storage medium including a computer program, and the computer program is executed by a processor to implement the steps of the drop monitoring method based on a radio frequency signal.

[0030] Compared with the prior art, the present application has at least the following beneficial effects:

[0031] The drop monitoring method based on a radio frequency signal stacks the received frames into a signal matrix, reduces the influence of a changing environment on the drop signal by subtracting the average change of the time window along the slow time, performs fast Fourier transform on the obtained signal, and then calculates the cross-correlation coefficients of the low frequency part and the high frequency part; the interval with the drop signal has a large coefficient change, and the interval without the drop signal does not have a large coefficient change, thereby positioning the signal interval where the drop is located, taking out the signal in the interval where the drop is located, and sending the signal into a neural network to predict the number of drops. The drop monitoring method can accurately locate the signal where the drop occurs in a relatively noisy environment, and can obtain a relatively accurate drop rate by extracting the periodic signal in the signal.

[0032] Further, using a commercial pulse wireless ultra-wideband (IR-UWB) radar to monitor drops without contact can reduce the time cost and labor cost caused by sensor pre-deployment, and can support simultaneous monitoring of multiple people at a long distance.

[0033] Further, using three one-dimensional convolution-based neural network modules can better extract droplet features, and applying a bidirectional LSTM module can better learn the time sequence relationship and extract the periodicity of the droplets to obtain more accurate droplet event estimation.

[0034] Further, using binary cross-entropy to train the neural network can better train the difference between the droplet event and the non-droplet event, and the binary cross-entropy function is also one of the classic loss functions.

[0035] Further, using the droplet rate correction method to infer the droplet rate can further correct the droplet event obtained by the neural network, reduce misjudgment and omission.

[0036] Further, using the droplet rate correction method to infer the droplet rate can further correct the droplet event obtained by the neural network, reduce misjudgment and omission.

[0037] It can be understood that the beneficial effects of the above-mentioned second aspect to the fourth aspect can be referred to the related description in the first aspect, which will not be repeated here.

[0038] In summary, the present application uses a commercial IR-UWB radar to monitor the droplets non-contact, which can be widely used in the medical field. At the same time, the present application does not need to pre-deploy sensors, has no requirements for light conditions, and does not involve privacy issues, avoiding the limitations of the existing droplet detection scheme, and realizing high-precision recovery of the droplet event.

[0039] The technical solutions of the present application will be further described in detail below by means of the drawings and examples. BRIEF DESCRIPTION OF DRAWINGS

[0040] Figure 1 It is the working principle diagram of IR-UWB radar;

[0041] Figure 2 It is the positioning droplet signal interval and preprocessing schematic diagram;

[0042] Figure 3 It is the droplet estimation network diagram;

[0043] Figure 4 Droplet estimation effect diagram;

[0044] Figure 5 Comparison diagram with other similar methods. DETAILED DESCRIPTION

[0045] Clearly, the described embodiments are only a part of the embodiments of the present application, but not all the embodiments. Based on the embodiments of the present application, all the other embodiments obtained by a person of ordinary skill in the art without creative effort are within the protection scope of the present application.

[0046] In the description of the present application, it should be understood that the terms "comprising" and "including" indicate the presence of the described features, integers, steps, operations, elements, and / or components, but do not exclude one or more other features, integers, steps, operations, elements, components, and / or groups thereof.

[0047] It should also be understood that the terms used in the present application specification are only for the purpose of describing particular embodiments and are not intended to limit the present application. As used in the present application specification and the appended claims, unless otherwise clearly indicated, the singular forms "a," "an," and "the" are intended to include the plural forms as well.

[0048] It should be further understood that the term "and / or" used in the present application specification and the appended claims means one or more of the associated listed items as well as all possible combinations thereof and includes these combinations, for example, A and / or B can mean A alone, A and B together, and B alone. In addition, the character " / " in this document generally represents an "or" relationship between the front and rear associated objects.

[0049] It should be understood that although the terms first, second, third, etc. can be used in the embodiments of the present application to describe preset ranges, etc., these preset ranges should not be limited to these terms. These terms are only used to distinguish the preset ranges from each other. For example, the first preset range can also be referred to as the second preset range, and similarly, the second preset range can also be referred to as the first preset range without departing from the scope of the embodiments of the present application.

[0050] Depending on the context, the word "if" as used herein can be interpreted to mean "when" or "while" or "in response to determining" or "in response to detecting". Similarly, the phrase "if determined" or "if detecting (a stated condition or event)" can be interpreted to mean "when determined" or "in response to determining" or "when detecting (a stated condition or event)" or "in response to detecting (a stated condition or event)", depending on the context.

[0051] Various structural diagrams according to the disclosed embodiments of the present application are shown in the drawings. These diagrams are not drawn to scale, in which certain details are exaggerated for clarity of presentation and may be omitted. The shapes of various regions, layers shown in the drawings and their relative sizes and positional relationships are merely exemplary, and in actuality may deviate due to manufacturing tolerances or technical limitations, and regions / layers with different shapes, sizes, relative positions can be additionally designed by those skilled in the art according to actual needs.

[0052] The present application provides a kind of based on radio frequency signal's drop liquid monitoring method, using commercial pulse wireless ultra-wideband (IR-UWB) radar is perceived to drop liquid event, locates the interval where drop liquid event is in signal matrix, designs a neural network to locate drop liquid, then uses drop liquid rate correction method to deduce drop liquid rate to realize to drop liquid event high precision recovery, can be widely applied in medical field.

[0053] The present application provides a kind of based on radio frequency signal's drop liquid monitoring method, including the following steps:

[0054] S1, using pulse wireless ultra-wideband radar receives signal, forms signal matrix y (t, i), subtracts average along the direction of t to signal matrix y (t, i), to eliminate the signal of environmental change, i.e.

[0055] Please refer to Figure 1 , t is the index of sending frame, i indicates distance interval, each interval corresponds to the distance resolution of radar;Change caused by moving object can be searched along the direction of t.

[0056] Please refer to Figure 2 (a), as shown in the elimination result, in addition to some false peak, the fluctuation caused by dripping water becomes more obvious;And compared with human activity and environmental change, dripping water signal is sharp and short.

[0057] S2, the signal obtained in step S1 is subjected to fast Fourier transform (FFT) along the direction of t, to obtain frequency spectrum signal;After FFT, the frequency spectrum of dripping water signal is very similar to that of periodic pulse signal, and others are not, then the frequency spectrum signal is divided into two parts, high frequency part (medium frequency and above) and low frequency part (below medium frequency). The cross-correlation coefficient curves of the frequency spectrum signals of high frequency band and low frequency band are calculated;

[0058] Please refer to Figure 2 (b), distance interval 2 contains dripping water signal, and distance interval 20 contains breathing.

[0059] S3, dividing the cross-correlation coefficient curve obtained in step S2 into several windows, summing the standard deviations of all windows, finding the distance interval corresponding to the cross-correlation coefficient curve with the largest sum of standard deviations, and the distance interval corresponding to the distance interval is the distance of the drop signal, taking the signal in the distance interval along the direction of t, denoted as y(t0, i);

[0060] Referring to Figure 2 (c), the distance interval with dripping has a larger coefficient change, while breathing does not.

[0061] Referring to Figure 2 (d), the standard deviation of the correlation coefficient of each distance interval is shown. The value of distance interval 2 is much larger than that of other distance intervals, indicating that the method of the application effectively locates the dripping signal in the signal matrix.

[0062] S4, inputting the signal y(t0, i) obtained in step S3 into the neural network to predict the position of the signal fluctuation peak caused by the drop;

[0063] First, generate the data set. It is found in the experiment that due to different reasons for generating the peak value, the fluctuation caused by the drop shows a consistent pattern in the I / Q component, while the peak value caused by other activities does not. The fluctuation caused by the drop always appears in the I component of the signal. The false peak introduced by the unknown factor is not the case. Therefore, in addition to the signal amplitude, its I and Q components are also recorded and input into the neural network.

[0064] Referring to Figure 3 , the neural network is composed of three convolutional blocks, a bidirectional long short-time neural network block, and a fully connected block. The kernel of the convolutional block is all 3, each convolutional block is composed of one-dimensional convolution, BatchNorm block, ReLU block stacked three times, the input channel is 3, and the output channel is 64; the input channel of the bidirectional LSTM convolutional block is 64, and the output channel is 32. The input channel of the fully connected block is 32, and the output channel is 1.

[0065] The loss function used by the neural network is binary cross-entropy, that is:

[0066]

[0067] Where r is the actual value of the peak position, '1' indicates a drop event, and '0' indicates no drop event, is the prediction result.

[0068] S5, inputting the peak position information obtained in step S4 into the drop rate correction method to obtain the drop rate, further eliminating the false peak value, and realizing the drop monitoring based on the radio frequency signal.

[0069] Specifically, a dropping rate correction method is designed to further eliminate spurious peaks. The basic idea of ​​the dropping rate correction method is to check whether all dropping intervals are within a reasonable range; if not, add or delete droppings accordingly. The steps of the dropping rate correction method are as follows:

[0070] S501, Calculate the time interval between two adjacent dripping events.

[0071] S502. Calculate the median of all drop intervals.

[0072] S503. Compare the interval between every two drop events obtained in step S501 with the median obtained in step S502. If the interval is greater than 1.1 times the median or less than 0.9 times the median, it is considered that there is an omission or misjudgment.

[0073] S504. If an omission occurs, i.e. the interval between two adjacent drops is greater than 1.1 times the median, then find the drop event with the highest probability in the interval between the two drops as the replacement drop event.

[0074] S505. If a misjudgment occurs, i.e. the interval between two adjacent drops is less than 0.9 times the median, then remove one of the drop events so that the intervals before and after it are both near the median (not greater than 1.1 times and not less than 0.9 times).

[0075] In another embodiment of the present invention, a drip monitoring system based on radio frequency signals is provided. This system can be used to implement the above-mentioned drip monitoring method based on radio frequency signals. Specifically, the drip monitoring system based on radio frequency signals includes an acquisition module, a transformation module, a positioning module, a training module, and a calibration module.

[0076] The acquisition module acquires signal data and forms a signal matrix, and subtracts the average value from the signal matrix along the t direction.

[0077] The transformation module performs a fast Fourier transform on the signal matrix of the acquisition module along the t direction after subtracting the average value to obtain the spectrum signal. The spectrum signal is divided into high-frequency and low-frequency components, and the cross-correlation curves of the high-frequency and low-frequency spectrum signals are calculated.

[0078] The positioning module divides the cross-correlation coefficient curve obtained by the transformation module into several windows, determines the distance interval corresponding to the cross-correlation coefficient curve with the largest sum of standard deviations of the windows, assigns the distance interval to the droplet signal distance, and extracts the signal within the distance interval in the time direction, denoted as y(t0, i).

[0079] The training module inputs the signal y(t0, i) obtained from the positioning module into the neural network to predict the peak position information of the signal fluctuation caused by the droplet.

[0080] The correction module determines the drop rate according to the signal fluctuation peak position information obtained by the training module, so as to realize the drop monitoring based on the radio frequency signal.

[0081] In another embodiment of the present application, a terminal device is provided, which comprises a processor and a memory, the memory is used to store a computer program, the computer program comprises program instructions, and the processor is used to execute the program instructions stored in the computer storage medium. The processor can be a central processing unit (CPU), and can also be other general-purpose processors, digital signal processors (DSP), application specific integrated circuits (ASIC), field-programmable gate arrays (FPGA) or other programmable logic devices, discrete gates or transistor logic devices, discrete hardware components, etc. The processor is the computing core and control core of the terminal, and is suitable for implementing one or more instructions, and is particularly suitable for loading and executing one or more instructions to implement a corresponding method flow or a corresponding function. The processor in the embodiment of the present application can be used for the operation of the drop monitoring method based on the radio frequency signal, which comprises:

[0082] Signal data is collected and a signal matrix is formed, and the average value is subtracted from the signal matrix along the t direction; the signal matrix with the average value subtracted along the t direction is subjected to fast Fourier transform to obtain a frequency spectrum signal, the frequency spectrum signal is divided into a high-frequency part and a low-frequency part, and the cross-correlation coefficient curve of the frequency spectrum signals of the high-frequency part and the low-frequency part is calculated; the cross-correlation coefficient curve is divided into a plurality of windows, the distance interval corresponding to the cross-correlation coefficient curve with the maximum sum of window standard deviations is determined, the distance interval is corresponded to the drop signal distance, the signal in the distance interval is taken out in the time direction and recorded as y(t0, i); the signal y(t0, i) is input into a neural network to predict the signal fluctuation peak position information caused by the drop; the drop rate is determined according to the signal fluctuation peak position information, and the drop monitoring based on the radio frequency signal is realized.

[0083] In another embodiment of the present application, the present application also provides a storage medium, specifically a computer readable storage medium (Memory), which is a memory device in a terminal device, used for storing programs and data. It can be understood that the computer readable storage medium herein can include a built-in storage medium in the terminal device, and of course can also include an extended storage medium supported by the terminal device. The computer readable storage medium provides a storage space, which stores an operating system of the terminal. In addition, one or more instructions suitable for being loaded and executed by the processor are also stored in the storage space, and the instructions can be one or more computer programs (including program codes). It should be noted that the computer readable storage medium herein can be a high-speed RAM memory or a non-volatile memory such as at least one disk memory.

[0084] The one or more instructions stored in the computer readable storage medium can be loaded and executed by the processor to implement the corresponding steps of the liquid drop monitoring method based on the radio frequency signal in the above embodiments; the one or more instructions in the computer readable storage medium are loaded and executed by the processor to perform the following steps:

[0085] The signal data is collected and a signal matrix is formed, and the average value of the signal matrix along the t direction is subtracted; the signal matrix along the t direction is subjected to fast Fourier transform to obtain a frequency spectrum signal, the frequency spectrum signal is divided into a high frequency part and a low frequency part, and the cross-correlation coefficient curve of the frequency spectrum signals of the high frequency part and the low frequency part is calculated; the cross-correlation coefficient curve is divided into a plurality of windows, the distance interval corresponding to the cross-correlation coefficient curve with the maximum sum of window standard deviations is determined, the distance interval is corresponded to a liquid drop signal distance, and the signal in the distance interval is taken out in the time direction and recorded as y(t0, i); the signal y(t0, i) is input into a neural network to predict the signal fluctuation peak position information caused by the liquid drop; the liquid drop rate is determined according to the signal fluctuation peak position information, and the liquid drop monitoring based on the radio frequency signal is realized.

[0086] To make the purpose, technical solutions and advantages of the embodiments of the present application clearer, the technical solutions in the embodiments of the present application will be described clearly and completely below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments. The components of the embodiments of the present application described and shown in the drawings herein can be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of the present application provided in the drawings is not intended to limit the scope of the claimed present application, but only represents selected embodiments of the present application. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative labor fall within the scope of protection of the present application.

[0087] The drop monitoring method based on radio frequency signals was tested in a laboratory.

[0088] The present application first evaluates the relative error for different drop rates. In practice, the infusion rate is a key indicator of the drug dosage. The most common infusion rate is 40-60 drops per minute.

[0089] Referring to Figure 4 The error rate of the present application is from 0.12% to 0.79% in the range of 20 drops per minute to 80 drops per minute, which is better than the label-based method DropMonitor (0.32% to 2.1%). When the drop rate is 10 drops per minute, the error is relatively high (1.67% uncorrected and 0.98% corrected). That is because when the drop rate is very low, the time interval becomes more irregular. And when the drop rate is high, the liquid level fluctuates more sharply, making the drop rate estimation prone to error.

[0090] The present application also tests in different environments:

[0091] Referring to Figure 5 The performance of the present application is better than DropMonitor, and the relative error is reduced from 0.77% to 1.02% to 0.35% to 0.54%; through the drop rate correction method proposed by the present application, the relative error can be further reduced to 0.32% to 0.5%, which indicates that the drop rate estimation of the present application is very accurate.

[0092] In summary, the drop monitoring method, system, medium and device based on radio frequency signals of the present application use commercial pulse radio ultra-wideband (IR-UWB) radar to realize non-contact monitoring of drops; at the same time, the present application does not need to pre-deploy sensors, has no requirements for light conditions, and does not involve privacy issues, avoiding the limitations of existing drop detection solutions, and realizing high-precision recovery of drop events.

[0093] Those skilled in the art will appreciate that embodiments of the present application can be provided as methods, systems, or computer program products. Therefore, the present application can take the form of an entirely hardware embodiment, an entirely software embodiment, or an embodiment combining software and hardware aspects. Moreover, the present application can take the form of a computer program product implemented on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROMs, optical storage, etc.) containing computer-usable program code.

[0094] The computer program instructions can also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer-implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart block or blocks. Figure 1 one or more flow or blocks. Figure 1 one or more flow or blocks.

[0095] These computer program instructions can also be stored in a computer readable memory that can direct a computer or other programmable data processing apparatus to function in a particular manner, such that the instructions stored in the computer readable memory produce an article of manufacture including instructions which implement the function specified in the flowchart block or blocks. Figure 1 one or more flow or blocks. Figure 1 one or more flow or blocks.

[0096] The computer program instructions can also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer-implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart block or blocks. Figure 1 one or more flow or blocks. Figure 1 one or more flow or blocks.

[0097] The above merely provides the technical idea of the present application, and cannot be used to limit the protection scope of the present application. Any modification made according to the technical idea of the present application, on the basis of the technical solutions, falls within the protection scope of the present application.

Claims

1. A method of drop monitoring based on radio frequency signals, characterized by, The method comprises the following steps: S1, collecting signal data and forming a signal matrix, and subtracting the average value along the t direction of the signal matrix; S2, performing fast Fourier transform on the signal matrix obtained by subtracting the average value along the t direction in step S1 to obtain a frequency spectrum signal, dividing the frequency spectrum signal into a high-frequency part and a low-frequency part, and calculating the cross-correlation coefficient curve of the frequency spectrum signals of the high-frequency part and the low-frequency part; S3, dividing the cross-correlation coefficient curve obtained in step S2 into a plurality of windows, determining the distance interval corresponding to the cross-correlation coefficient curve with the maximum sum of window standard deviations, corresponding the distance interval to a drop signal distance, and taking out the signals in the distance interval in the time direction and recording them as y(t0, i); S4, inputting the signal y(t0, i) obtained in step S3 into a neural network to predict the signal fluctuation peak position information caused by the drop; S5, determining the drop rate according to the signal fluctuation peak position information obtained in step S4 to realize drop monitoring based on the radio frequency signal, and checking whether all drop intervals are within the set range by using a drop rate correction method to determine the drop rate, and the steps of the drop rate correction method are as follows: respectively calculating the interval time between two adjacent drop events and the median of all drop interval times, comparing the interval time between each two drop events with the median, determining that the adjacent two drop intervals are missed when the interval time is greater than 1.1 times the median, and taking the drop event with the maximum probability in the interval between the adjacent two drops as a supplement; when the interval time between the adjacent two drops is less than 0.9 times the median, it is determined that the adjacent two drops are misjudged, and one of the adjacent two drops is removed so that the intervals of the remaining drops are at the median.

2. The radio frequency signal based drop monitoring method of claim 1, wherein, In step S1, an IR-UWB radar is used to collect signal data.

3. The radio frequency signal based drop monitoring method of claim 1, wherein, In step S4, the neural network comprises three residual convolution blocks, one bidirectional long short-time neural network block, and one fully connected block.

4. The radio frequency signal based drop monitoring method of claim 3, wherein, The loss function of the neural network is binary cross entropy.

5. The radio frequency signal based drop monitoring method of claim 4, wherein, Loss function is: wherein, is the actual value of the peak position, is the prediction result.

6. A drop monitoring system based on radio frequency signals, characterized in that It comprises: a collection module for collecting signal data and forming a signal matrix, and subtracting the average value along the t direction of the signal matrix; a transformation module for performing fast Fourier transform on the signal matrix obtained by subtracting the average value along the t direction in the collection module to obtain a frequency spectrum signal, dividing the frequency spectrum signal into a high-frequency part and a low-frequency part, and calculating the cross-correlation coefficient curve of the frequency spectrum signals of the high-frequency part and the low-frequency part; a positioning module for dividing the cross-correlation coefficient curve obtained by the transformation module into a plurality of windows, determining the distance interval corresponding to the cross-correlation coefficient curve with the maximum sum of window standard deviations, corresponding the distance interval to a drop signal distance, and taking out the signals in the distance interval in the time direction and recording them as y(t0, i); a training module for inputting the signal y(t0, i) obtained by the positioning module into a neural network to predict the signal fluctuation peak position information caused by the drop; a correction module for determining the drop rate according to the signal fluctuation peak position information obtained by the training module to realize drop monitoring based on the radio frequency signal, and checking whether all drop intervals are within the set range by using a drop rate correction method to determine the drop rate, and the steps of the drop rate correction method are as follows: The interval time between two adjacent drop events is calculated respectively, and the median of all drop interval times is calculated, then the interval time of each two drop events is compared with the median, when the interval time of two adjacent drops is greater than 1.1 times of the median, it is determined as missing, the drop event with the maximum probability in the interval between the two adjacent drops is supplemented; when the interval time of two adjacent drops is less than 0.9 times of the median, it is determined as misjudgment, one drop event in the two adjacent drops is removed, and the interval of the remaining drops is at the median.

7. A computer-readable storage medium storing one or more programs, the one or more programs comprising instructions that when executed by a computer cause the computer to perform a method of any of claims 1-6. The one or more programs include instructions that, when executed by a computing device, cause the computing device to perform any of the methods of claims 1-5.

8. A computing device, comprising: comprise: one or more processors, memory, and one or more programs, wherein the one or more programs are stored in the memory and configured to be executed by the one or more processors, the one or more programs including instructions for performing any of the methods of claims 1-5.

Images