A control method and system for continuous infusion and drug replacement based on a multi-connection drug liquid tank

Abstract

The present application relates to a kind of control method and system based on multi-liquid medicine tank's continuous infusion drug change, the method includes:U1.It is hung on bracket and carries out infusion, based on the image data information of camera on bracket real-time acquisition medicine tank, based on the image data information of camera on bracket real-time acquisition medicine drop, based on the quantity data information of laser counting sensor on bracket real-time acquisition medicine drop drop;U2.Based on the image data information of the medicine tank and the image data information of the medicine drop, using improved Shi-Tomasi corner detection algorithm, the image of medicine tank and medicine drop is extracted, and the image feature data information of medicine tank and the image feature data information of medicine drop are obtained.The present application not only can accurately control the infusion process of medicine tank, and the whole process, without manual drug change, improve the efficiency and stability of infusion, to improve the use experience of patient.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of multi-connection medicine liquid tank drug replacement, in particular to a control method and system for continuous infusion drug replacement based on a multi-connection medicine liquid tank. BACKGROUND

[0002] With the continuous development of automation technology, it is constantly affecting the rapid development of various industries, among them, the medical industry is also constantly updating and developing, and innovation for each technical point in the industry is the direction of our research and development.

[0003] In the prior art, a patent (application number: 202110549422.4) discloses a precise, safe and convenient infusion drug replacement method, which comprises the following steps: S01: information collection, before infusion, medical staff reads the medicine information by using an information collection device, and marks the total volume V of the medicine; S03: infusion time determination, the infusion control robot collects the current infusion speed v, and sets the control variable K, during the infusion process, the system calculates the medicine infusion stop time T through the above parameters, the medicine infusion stop time T = total volume V of the medicine * control variable K / infusion speed v. The accuracy of the drug replacement in the scheme is not high, and the medicine liquid bottle still needs to be replaced manually, which not only wastes time, but also reduces the efficiency of medicine liquid delivery. SUMMARY

[0004] In view of the deficiencies of the above prior art, the present application provides a control method and system for continuous infusion drug replacement based on a multi-connection medicine liquid tank, which can accurately control the infusion process of the medicine liquid tank, and the whole process does not need manual drug replacement, improves the efficiency and stability of infusion, and thus improves the use experience of patients.

[0005] In order to achieve the above-mentioned purpose and other related purposes, the technical scheme provided by the present application is as follows:

[0006] A control method for continuous infusion drug replacement based on a multi-connection medicine liquid tank, the method comprising:

[0007] U1. suspending the multi-connection medicine liquid tank on a support for infusion, acquiring image data information of the medicine liquid tank based on a camera on the support, acquiring image data information of the medicine liquid drops based on the camera on the support, and acquiring quantity data information of the falling medicine liquid drops based on a laser counting sensor on the support;

[0008] U2. based on the image data information of the medicine liquid tank and the image data information of the medicine liquid drops, using an improved Shi-Tomasi corner detection algorithm to extract features of the images of the medicine liquid tank and the medicine liquid drops, to obtain image feature data information of the medicine liquid tank and image feature data information of the medicine liquid drops;

[0009] U3. Based on the image feature data information of the medicine liquid tank and the image feature data information of the medicine liquid drops, a recurrent neural network prediction model based on the Spearman correlation coefficient is constructed to predict the number of medicine liquid drops in the medicine liquid tank, and the number data information of the medicine liquid drops in the medicine liquid tank after prediction is obtained;

[0010] U4. Based on the number data information of the medicine liquid drops in the medicine liquid tank after prediction and the number data information of the falling medicine liquid drops, a continuous infusion and drug replacement control function G of the multi-connected medicine liquid tank is established to control and adjust the continuous infusion and drug replacement of the medicine liquid tank, and the control data information of the continuous infusion and drug replacement of the multi-connected medicine liquid tank is obtained.

[0011] Further, the continuous infusion and drug replacement control function G of the multi-connected medicine liquid tank is,

[0012]

[0013] Wherein, x is the number data information of the medicine liquid drops in the medicine liquid tank after prediction, y is the number data information of the falling medicine liquid drops, α1 is the first control factor of the multi-connected medicine liquid tank, α2 is the second control factor of the multi-connected medicine liquid tank, and α3 is the third control factor of the multi-connected medicine liquid tank.

[0014] Further, the first control factor α1 of the multi-connected medicine liquid tank is,

[0015]

[0016] The second control factor α2 of the multi-connected medicine liquid tank is,

[0017]

[0018] The third control factor α3 of the multi-connected medicine liquid tank is,

[0019]

[0020] Wherein, x is the number data information of the medicine liquid drops in the medicine liquid tank after prediction, y is the number data information of the falling medicine liquid drops.

[0021] Further, in step U3, the construction of the recurrent neural network prediction model based on the Spearman correlation coefficient to predict the number of medicine liquid drops in the medicine liquid tank comprises:

[0022] U31. Based on the image feature data information of the medicine liquid tank and the image feature data information of the medicine liquid drops, a Spearman correlation coefficient function H of the medicine liquid tank and the medicine liquid drops is established,

[0023]

[0024] Wherein, a iis the image feature data information of the liquid medicine tank, n is the sample capacity, b i is the image feature data information of the liquid medicine drop, f is the order function of the image feature of the liquid medicine tank, g is the order function of the image feature of the liquid medicine drop, the correlation between the image features of the liquid medicine tank and the liquid medicine drop is characterized, and the data information of the Spearman correlation coefficient of the liquid medicine tank and the liquid medicine drop is obtained;

[0025] U32. The data information of the Spearman correlation coefficient of the liquid medicine tank and the liquid medicine drop is input into a recurrent neural network prediction model for training and learning to determine a prediction function W of the liquid medicine tank drop number,

[0026]

[0027] Wherein, c is the data information of the Spearman correlation coefficient of the liquid medicine tank and the liquid medicine drop, and a1, a2 and a3 are learning factors of the prediction model, and a trained recurrent neural network model is obtained;

[0028] U33. Based on the trained recurrent neural network model, the image feature data information of the liquid medicine tank and the image feature data information of the liquid medicine drop are input to predict the number of liquid medicine drops in the liquid medicine tank, and the number data information of the liquid medicine drops in the liquid medicine tank after prediction is obtained.

[0029] Further, the order function f of the image feature of the liquid medicine tank is

[0030]

[0031] The order function g of the image feature of the liquid medicine drop is

[0032]

[0033] Wherein, a i is the image feature data information of the liquid medicine tank, n is the sample capacity, b i is the image feature data information of the liquid medicine drop.

[0034] Further, the constraint conditions of the learning factors a1, a2 and a3 of the prediction model are

[0035] Further, in step U2, the feature extraction of the liquid medicine tank and the liquid medicine drop image by using the improved Shi-Tomasi corner detection algorithm includes:

[0036] U21. Based on the image data information of the liquid medicine tank and the image data information of the liquid medicine drop, the image pixel matrix of the liquid medicine tank and the image pixel matrix of the liquid medicine drop are constructed, and the image pixel matrix data information of the liquid medicine tank and the liquid medicine drop is obtained;

[0037] U22. Based on the image pixel matrix data information of the liquid medicine tank and the liquid medicine drop, an image pixel corner point detection function R1 of the liquid medicine tank and an image pixel corner point detection function R2 of the liquid medicine drop are established,

[0038]

[0039] wherein p is the image pixel matrix data information of the liquid medicine tank, q is the image pixel matrix data information of the liquid medicine drop, μ1, μ2 and μ3 are the image pixel detection factors of the liquid medicine tank, ρ1, ρ2 and ρ3 are the image pixel detection factors of the liquid medicine drop;

[0040] U23. Based on the image pixel corner point detection function R1 of the liquid medicine tank and the image pixel corner point detection function R2 of the liquid medicine drop, the images of the liquid medicine tank and the liquid medicine drop are subjected to feature extraction to obtain image feature data information of the liquid medicine tank and image feature data information of the liquid medicine drop.

[0041] Further, in step U23, the feature extraction of the images of the liquid medicine tank and the liquid medicine drop is to obtain the image pixel corner point detection value of the liquid medicine tank and the image pixel corner point detection value of the liquid medicine drop according to the image pixel corner point detection function R1 of the liquid medicine tank and the image pixel corner point detection function R2 of the liquid medicine drop, and a preset threshold value is set, if the image pixel corner point detection value of the liquid medicine tank is less than the preset threshold value, it is rejected, if the image pixel corner point detection value of the liquid medicine tank is greater than the preset threshold value, it is retained, if the image pixel corner point detection value of the liquid medicine drop is less than the preset threshold value, it is rejected, if the image pixel corner point detection value of the liquid medicine drop is greater than the preset threshold value, it is retained.

[0042] Further, in step U4, the control and adjustment of the continuous infusion and drug replacement of the liquid medicine tank is to obtain the number of liquid medicine drops of each liquid medicine tank in the multi-connection liquid medicine tank according to the continuous infusion and drug replacement control function G of the multi-connection liquid medicine tank, and set a liquid medicine drop threshold value, if the number of liquid medicine drops of the liquid medicine tank is less than the liquid medicine drop threshold value, another liquid medicine tank needs to be replaced for infusion, if the number of liquid medicine drops of the liquid medicine tank is greater than the liquid medicine drop threshold value, the infusion is normally carried out.

[0043] In order to achieve the above-mentioned purposes and other related purposes, the present application also provides a continuous infusion and drug replacement control system based on multi-connection liquid medicine tank, comprising a computer device which is programmed or configured to perform the steps of any one of the continuous infusion and drug replacement methods based on multi-connection liquid medicine tank.

[0044] The present application has the following positive effects:

[0045] 1.The method can not only accurately predict the infusion process of the liquid medicine tank, ensure the smoothness of the liquid medicine tank replacement process, improve the stability of the continuous infusion and drug replacement process, but also can be adjusted adaptively for different liquid medicine tanks and liquid medicines, without the need to reset parameters, improve the convenience and adaptability, and has strong robustness.

[0046] 2.The method can control and adjust the liquid medicine tank of the continuous infusion and drug replacement, not only the whole process, without manual drug replacement, improve the efficiency and stability of infusion, so as to improve the patient's use experience, and further improve the smoothness of the continuous infusion and drug replacement. BRIEF DESCRIPTION OF DRAWINGS

[0047] Figure 1 It is a method flowchart of the present application.

[0048] Figure 2 It is a flowchart of the improved Shi-Tomasi corner detection algorithm of the present application.

[0049] Figure 3 It is a flowchart of constructing a recurrent neural network prediction model based on the Spearman correlation coefficient of the present application.

[0050] Figure 4 It is a structural schematic diagram of the present application.

[0051] Explanation of figure marks: 1-bracket, 2-camera, 3-laser counting sensor, 4-liquid medicine tank, 5-liquid medicine drop. DETAILED DESCRIPTION

[0052] The exemplary embodiments of the present disclosure are described below in conjunction with the accompanying drawings, including various details of the embodiments of the present disclosure to help understanding, which should be considered only as exemplary. Therefore, those skilled in the art should recognize that various changes and modifications can be made to the embodiments described herein without departing from the scope and spirit of the present disclosure. Also, for the sake of clarity and conciseness, the description below omits the description of well-known functions and structures.

[0053] Embodiment 1: as shown in Figure 1 or Figure 4 A control method for continuous infusion and drug replacement based on multi-liquid medicine tank, the method comprises:

[0054] U1. The multi-connection medicine liquid tank 4 is hung on the support 1 for infusion, the image data information of the medicine liquid tank 4 is acquired in real time based on the camera 2 on the support 1, the image data information of the medicine liquid drop 5 is acquired in real time based on the camera 2 on the support 1, and the quantity data information of the medicine liquid drop 5 falling is acquired in real time based on the laser counting sensor 3 on the support 1;

[0055] U2. Based on the image data information of the medicine liquid tank and the image data information of the medicine liquid drop, the improved Shi-Tomasi corner detection algorithm is used to extract the features of the images of the medicine liquid tank and the medicine liquid drop, and the image feature data information of the medicine liquid tank and the image feature data information of the medicine liquid drop are obtained;

[0056] U3. Based on the image feature data information of the medicine liquid tank and the image feature data information of the medicine liquid drop, a recurrent neural network prediction model based on the Spearman correlation coefficient is constructed to predict the quantity of the medicine liquid drop in the medicine liquid tank, and the quantity data information of the medicine liquid drop in the medicine liquid tank after prediction is obtained;

[0057] U4. Based on the quantity data information of the medicine liquid drop in the medicine liquid tank after prediction and the quantity data information of the falling medicine liquid drop, a multi-connection medicine liquid tank continuous infusion and drug replacement control function G is established to control and adjust the medicine liquid tank for continuous infusion and drug replacement, and the control data information of the continuous infusion and drug replacement of the multi-connection medicine liquid tank is obtained.

[0058] In this embodiment, the continuous infusion and drug replacement control function G of the multi-connection medicine liquid tank is,

[0059]

[0060] Wherein, x is the quantity data information of the medicine liquid drop in the medicine liquid tank after prediction, y is the quantity data information of the falling medicine liquid drop, α1 is the first control factor of the multi-connection medicine liquid tank, α2 is the second control factor of the multi-connection medicine liquid tank, and α3 is the third control factor of the multi-connection medicine liquid tank.

[0061] In this embodiment, the first control factor α1 of the multi-connection medicine liquid tank is,

[0062]

[0063] The second control factor α2 of the multi-connection medicine liquid tank is,

[0064]

[0065] The third control factor α3 of the multi-connection medicine liquid tank is,

[0066]

[0067] Where x represents the predicted number of medicine droplets in the medicine container, and y represents the number of dripping medicine droplets.

[0068] In this embodiment, as Figure 3 As shown, in step U3, the construction of a recurrent neural network prediction model based on Spearman correlation coefficient to predict the number of drug droplets in the drug container includes:

[0069] U31. Based on the image feature data of the medicine container and the image feature data of the medicine droplets, a Spearman correlation coefficient function H between the medicine container and the medicine droplets is established.

[0070]

[0071] Among them, a i The image feature data of the medicine container, where n is the sample size and b is the sample size. i The image feature data of the liquid medicine droplet is given, f is the position function of the image feature of the liquid medicine container, and g is the position function of the image feature of the liquid medicine droplet. The correlation between the image features of the liquid medicine container and the liquid medicine droplet is characterized, and the Spearman correlation coefficient data of the liquid medicine container and the liquid medicine droplet is obtained.

[0072] U32. Input the Spearman correlation coefficient data between the medicine container and the medicine droplets into the recurrent neural network prediction model for training and learning, and determine the prediction function W for the number of medicine droplets in the container.

[0073]

[0074] Where c represents the Spearman correlation coefficient between the liquid container and the liquid droplets, and α1, α2 and α3 are the learning factors of the prediction model, resulting in a trained recurrent neural network model.

[0075] U33. Based on the trained recurrent neural network model, input the image feature data of the medicine container and the image feature data of the medicine droplets, predict the number of medicine droplets in the medicine container, and obtain the predicted number of medicine droplets in the medicine container.

[0076] In this embodiment, the position function f of the image features of the medicine container is,

[0077]

[0078] The position function g of the image features of the drug droplet is,

[0079]

[0080] Among them, a i b is the image feature data information of the medicine tank.i The image feature data information of the liquid medicine drops.

[0081] In this embodiment, the constraint conditions of the learning factors a1, a2 and a3 of the prediction model are,

[0082]

[0083] Embodiment 2: Based on the control method of the multi-connection liquid medicine tank continuous infusion and drug replacement in embodiment 1, the present application is further described and explained as follows.

[0084] As shown in Figure 1 or Figure 4 A control method of multi-connection liquid medicine tank continuous infusion and drug replacement, the method comprising:

[0085] U1. The multi-connection liquid medicine tank 4 is hung on the support 1 for infusion, the image data information of the liquid medicine tank 4 is obtained in real time based on the camera 2 on the support 1, the image data information of the liquid medicine drops 5 is obtained in real time based on the camera 2 on the support 1, and the quantity data information of the falling liquid medicine drops 5 is obtained in real time based on the laser counting sensor 3 on the support 1;

[0086] U2. Based on the image data information of the liquid medicine tank and the image data information of the liquid medicine drops, the improved Shi-Tomasi corner detection algorithm is used to extract the features of the images of the liquid medicine tank and the liquid medicine drops, to obtain the image feature data information of the liquid medicine tank and the image feature data information of the liquid medicine drops;

[0087] U3. Based on the image feature data information of the liquid medicine tank and the image feature data information of the liquid medicine drops, a recurrent neural network prediction model based on the Spearman correlation coefficient is constructed to predict the quantity of the liquid medicine drops in the liquid medicine tank, to obtain the quantity data information of the liquid medicine drops in the liquid medicine tank after prediction;

[0088] U4. Based on the quantity data information of the liquid medicine drops in the liquid medicine tank after prediction and the quantity data information of the falling liquid medicine drops, a multi-connection liquid medicine tank continuous infusion and drug replacement control function G is established to control and adjust the liquid medicine tank for continuous infusion and drug replacement, to obtain the control data information of the multi-connection liquid medicine tank continuous infusion and drug replacement.

[0089] In this embodiment, as shown in Figure 2 in step U2, the improved Shi-Tomasi corner detection algorithm is used to extract the features of the images of the liquid medicine tank and the liquid medicine drops, comprising:

[0090] U21. Based on the image data information of the liquid medicine tank and the image data information of the liquid medicine drops, the image pixel matrix of the liquid medicine tank and the image pixel matrix of the liquid medicine drops are constructed, to obtain the image pixel matrix data information of the liquid medicine tank and the liquid medicine drops;

[0091] U22. Based on the image pixel matrix data information of the liquid medicine tank and the liquid medicine drop, an angle point detection function R1 of the image pixel of the liquid medicine tank and an angle point detection function R2 of the image pixel of the liquid medicine drop are established,

[0092]

[0093] wherein p is the image pixel matrix data information of the liquid medicine tank, q is the image pixel matrix data information of the liquid medicine drop, μ1, μ2 and μ3 are the image pixel detection factors of the liquid medicine tank, and ρ1, ρ2 and ρ3 are the image pixel detection factors of the liquid medicine drop;

[0094] U23. Based on the angle point detection function R1 of the image pixel of the liquid medicine tank and the angle point detection function R2 of the image pixel of the liquid medicine drop, the images of the liquid medicine tank and the liquid medicine drop are subjected to feature extraction to obtain image feature data information of the liquid medicine tank and image feature data information of the liquid medicine drop.

[0095] In the present embodiment, in step U23, the feature extraction of the images of the liquid medicine tank and the liquid medicine drop is to obtain the angle point detection values of the image pixels of the liquid medicine tank and the angle point detection values of the image pixels of the liquid medicine drop according to the angle point detection function R1 of the image pixel of the liquid medicine tank and the angle point detection function R2 of the image pixel of the liquid medicine drop, and a preset threshold value is set, and if the angle point detection value of the image pixel of the liquid medicine tank is less than the preset threshold value, it is removed, if the angle point detection value of the image pixel of the liquid medicine tank is greater than the preset threshold value, it is retained, if the angle point detection value of the image pixel of the liquid medicine drop is less than the preset threshold value, it is removed, if the angle point detection value of the image pixel of the liquid medicine drop is greater than the preset threshold value, it is retained.

[0096] In the present embodiment, in step U4, the control and adjustment of the continuous infusion and drug replacement of the liquid medicine tank is to obtain the number of liquid medicine drops of each liquid medicine tank in the multi-connection liquid medicine tank according to the continuous infusion and drug replacement control function G of the multi-connection liquid medicine tank, set a liquid medicine drop threshold value, if the number of liquid medicine drops of the liquid medicine tank is less than the liquid medicine drop threshold value, another liquid medicine tank needs to be replaced for infusion, if the number of liquid medicine drops of the liquid medicine tank is greater than the liquid medicine drop threshold value, the infusion is normally carried out.

[0097] In the present embodiment, the present application provides a control system for continuous infusion and drug replacement based on multi-connection liquid medicine tank, comprising a computer device programmed or configured to perform the steps of any one of the methods for continuous infusion and drug replacement based on multi-connection liquid medicine tank.

[0098] In the present embodiment, the present application provides a computer readable storage medium having stored thereon a computer program programmed or configured to perform any one of the improved Yolov5 pedestrian detection methods.

[0099] Any reference to storage, memory, database or other medium herein can include non-volatile and / or volatile storage. Non-volatile storage can include read-only memory (ROM), programmable ROM (PROM), electronically programmable ROM (EPROM), or electrically erasable programmable ROM (EEPROM). Volatile storage can include random-access memory (RAM). By way of illustration, and not limitation, RAM is available in many forms such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDR SDRAM), enhanced SDRAM (ESDRAM), Synchlink DRAM (SLDRAM), Rambus direct RAM (RDRAM), direct Rambus dynamic RAM (DRDRAM), and Rambus dynamic RAM (RDRAM). The RAM can also include a basic-oxide-of-silicon (BOS) memory.

[0100] In summary, the present application can not only accurately control the infusion process of the medicine tank, but also improve the efficiency and stability of the infusion without manual medicine replacement, thereby improving the use experience of the patient.

[0101] The above detailed description does not constitute a limitation on the protection scope of the present disclosure. Those skilled in the art should understand that various modifications, combinations, sub-combinations, and substitutions can be made according to design requirements and other factors. Any modifications, equivalent replacements, and improvements made within the spirit and principles of the present disclosure shall be included in the protection scope of the present disclosure.

Claims

1. A control method for continuous drug delivery and replacement based on multi-unit drug tanks, characterized in that, The method includes: U1. The multi-unit drug infusion tank is suspended on the bracket for infusion. The camera on the bracket acquires image data of the drug infusion tank in real time, the camera on the bracket acquires image data of the drug droplets in real time, and the laser counting sensor on the bracket acquires the number of dripping drug droplets in real time. U2. Based on the image data information of the liquid medicine container and the image data information of the liquid medicine droplets, the improved Shi-Tomasi corner detection algorithm is used to extract features from the images of the liquid medicine container and the liquid medicine droplets to obtain the image feature data information of the liquid medicine container and the image feature data information of the liquid medicine droplets; U3. Based on the image feature data of the liquid medicine container and the image feature data of the liquid medicine droplets, a recurrent neural network prediction model based on Spearman correlation coefficient is constructed to predict the number of liquid medicine droplets in the liquid medicine container, and the predicted number of liquid medicine droplets in the liquid medicine container is obtained. U4. Based on the predicted number of drug droplets in the drug tank and the number of dripping drug droplets, establish a continuous drug exchange control function G for the multi-unit drug tank, control and adjust the continuous drug exchange drug tank, and obtain the control data information for the continuous drug exchange drug exchange of the multi-unit drug tank. The continuous transport and drug replacement control function G of the multi-unit drug tank is: ， Where x represents the predicted number of drug droplets in the drug tank, y represents the number of dripping drug droplets, α1 represents the first control factor of the multi-unit drug tank, α2 represents the second control factor of the multi-unit drug tank, and α3 represents the third control factor of the multi-unit drug tank. The first control factor α1 of the multi-unit drug tank is... ， The second control factor α2 of the multi-unit drug tank is: ， The third control factor α3 of the multi-unit drug tank is, ， Where x represents the predicted number of medicine droplets in the medicine container, and y represents the number of dripping medicine droplets.

2. The control method for continuous drug delivery and replacement based on multi-unit drug tanks according to claim 1, characterized in that, In step U3, the construction of a recurrent neural network prediction model based on Spearman correlation coefficient to predict the number of drug droplets in the drug container includes: U31. Based on the image feature data of the medicine container and the image feature data of the medicine droplets, a Spearman correlation coefficient function H between the medicine container and the medicine droplets is established. ， Among them, a i The image feature data of the medicine container, where n is the sample size and b is the sample size. i The image feature data of the liquid medicine droplet is given, f is the position function of the image feature of the liquid medicine container, and g is the position function of the image feature of the liquid medicine droplet. The correlation between the image features of the liquid medicine container and the liquid medicine droplet is characterized, and the Spearman correlation coefficient data of the liquid medicine container and the liquid medicine droplet is obtained. U32. Input the Spearman correlation coefficient data between the medicine container and the medicine droplets into the recurrent neural network prediction model for training and learning, and determine the prediction function W for the number of medicine droplets in the container. ， Where c represents the Spearman correlation coefficient between the liquid container and the liquid droplets, and β1, β2 and β3 are the learning factors of the prediction model, resulting in a well-trained recurrent neural network model. U33. Based on the trained recurrent neural network model, input the image feature data of the medicine container and the image feature data of the medicine droplets, predict the number of medicine droplets in the medicine container, and obtain the predicted number of medicine droplets in the medicine container.

3. The control method for continuous drug delivery and replacement based on multi-unit drug tanks according to claim 2, characterized in that: The position function f of the image features of the medicine container is, ， The position function g of the image features of the drug droplet is, ， Among them, a i b is the image feature data information of the medicine tank. i This refers to the image feature data information of the drug droplets.

4. The control method for continuous drug delivery and replacement based on multi-unit drug tanks according to claim 2, characterized in that: The constraints for the learning factors β1, β2, and β3 of the prediction model are as follows: 。 5. The control method for continuous drug delivery and replacement based on multi-unit drug tanks according to claim 1, characterized in that, In step U2, the feature extraction of the images of the medicine container and medicine droplets using the improved Shi-Tomasi corner detection algorithm includes: U21. Based on the image data information of the liquid medicine container and the image data information of the liquid medicine droplets, construct the image pixel matrix of the liquid medicine container and the image pixel matrix of the liquid medicine droplets to obtain the image pixel matrix data information of the liquid medicine container and the liquid medicine droplets. U22. Based on the image pixel matrix data of the medicine container and the medicine droplets, establish a corner detection function R1 for the image pixels of the medicine container and a corner detection function R2 for the image pixels of the medicine droplets. ， ， Where p is the image pixel matrix data information of the liquid container, q is the image pixel matrix data information of the liquid droplet, µ1, µ2 and µ3 are the image pixel detection factors of the liquid container, and ρ1, ρ2 and ρ3 are the image pixel detection factors of the liquid droplet. U23. Based on the corner detection function R1 of the image pixels of the liquid container and the corner detection function R2 of the image pixels of the liquid droplet, feature extraction is performed on the images of the liquid container and the liquid droplet to obtain the image feature data information of the liquid container and the image feature data information of the liquid droplet.

6. The control method for continuous drug delivery and replacement based on multi-unit drug tanks according to claim 5, characterized in that, In step U23, the feature extraction of the images of the liquid container and the liquid droplets involves obtaining the corner detection values ​​of the image pixels of the liquid container and the image pixels of the liquid droplets based on the corner detection function R1 of the image pixels of the liquid container and the corner detection function R2 of the image pixels of the liquid droplets, and setting a preset threshold. If the corner detection value of the image pixels of the liquid container is less than the preset threshold, it is discarded; if the corner detection value of the image pixels of the liquid container is greater than the preset threshold, it is retained.

7. The control method for continuous drug delivery and replacement based on multi-unit drug tanks according to claim 1, characterized in that, In step U4, the control and adjustment of the continuous infusion and medication replacement tanks is to obtain the number of drug drops in each of the multi-tank infusion tanks according to the continuous infusion and medication replacement control function G of the multi-tank infusion tanks, set a drug drop count threshold, and if the number of drug drops in a tank is less than the drug drop count threshold, it needs to be replaced with another drug tank for infusion, and if the number of drug drops in a tank is greater than the drug drop count threshold, the infusion is performed normally.

8. A control system for continuous drug delivery and replacement based on multi-unit drug tanks, comprising computer equipment, characterized in that, The computer device is programmed or configured to perform the steps of the control method for continuous drug delivery and replacement based on a multi-tank system as described in any one of claims 1 to 7.

Images