A method for collecting blood components

Abstract

The present invention discloses a blood component collection method, which relates to the field of intelligent blood collection control. By using the Lambert diffuse reflection model and the Phong specular reflection model, the present invention effectively separates different light components, and further extracts the time-frequency characteristics of the blood layer through short-time Fourier transform, enhancing the accuracy of component recognition. In addition, the present invention uses a hierarchical perception radial basis function neural network for classification, mapping the optical features to specific blood component categories to achieve simultaneous recognition and separation of multiple components; and dynamically adjusts the blood pump flow rate through a PID controller to ensure the stability of the hierarchical structure and improve the real-time performance and controllability of the separation process.

Description

Technical Field

[0001] The present invention relates to the field of intelligent blood collection control, and specifically to a method for collecting blood components. Background Art

[0002] In the field of blood component collection, existing technologies mainly rely on centrifugal separation and membrane separation. The centrifugal separation method uses a high-speed rotating centrifuge to layer different components in whole blood according to density differences. Common components include red blood cells, white blood cells, platelets, and plasma. The core of the centrifugal separation method is to use centrifugal force to layer blood components according to density and then collect them through different collection ports. The membrane separation method uses the selective permeability of a semi-permeable membrane to separate different components in blood. The key to the membrane separation method lies in the selection of membrane materials and the design of membrane pore sizes to achieve effective separation of different components. In addition, there are also some technologies based on optical detection that identify and separate components by detecting the optical properties of blood components. These technologies usually combine methods such as spectral analysis, light scattering, and reflected light intensity detection to improve the accuracy and efficiency of separation.

[0003] Although existing blood component collection technologies meet clinical needs to a certain extent, there are still some significant defects, which are exactly the technical problems that this solution aims to solve. First of all, although the centrifugal separation method can effectively separate blood components, its separation process depends on high-speed rotation, which easily causes mechanical damage to blood components, especially has a greater impact on fragile components such as platelets and white blood cells. In addition, the separation efficiency of the centrifugal separation method is limited by the rotation speed and separation time of the centrifuge, and it is difficult to achieve real-time and dynamic separation control, resulting in a lack of flexibility and adaptability in the separation process. Although the membrane separation method avoids the problem of mechanical damage, the selection of membrane materials and the design of membrane pore sizes limit its scope of application, and membrane fouling and blockage problems are difficult to avoid, resulting in a decrease in separation efficiency and an increase in maintenance costs. Although optical detection technologies have advantages in component identification, their detection accuracy is limited by the stability of the light source and interference from the detection environment, and it is difficult to achieve high-precision component separation. In addition, when dealing with complex blood samples, existing technologies often have difficulty in simultaneously separating and identifying multiple components, resulting in low separation efficiency and complex operations. Existing technologies also lack the ability to monitor and dynamically adjust the separation process in real time, and cannot adjust separation parameters in a timely manner according to changes in the separation state, resulting in unstable separation effects. These technical problems limit the further development and application of blood component collection technologies. Summary of the Invention

[0004] The present invention proposes a method for collecting blood components. By combining optical signal analysis and a hybrid neural network model, and finally performing dynamic adjustment through PID control, high-precision separation of blood components is achieved.

[0005] Among them, a blood component collection method includes the following steps:

[0006] S1. Pump whole blood into the separation chamber through a whole blood pump, and rotate it at high speed by a motor, so that the whole blood in the separation chamber is stratified according to the density of blood components. At the same time, emit light from a light source to irradiate different blood component layers in the separation chamber, and collect the intensity of reflected light in a time series;

[0007] S2. Calculate the time change rate of the reflected light intensity and the second-order change rate of the reflected light intensity according to the collected reflected light intensity in the time series;

[0008] S3. Separate different light components through the Lambert diffuse reflection model and the Phong specular reflection model to obtain the optical characteristics of the reflected light intensity; extract the time-frequency characteristics of the blood layer through short-time Fourier transform;

[0009] S4. Construct the input feature vector of the neural network, input the input vector into the hierarchical perception radial basis function neural network for classification, and map the optical characteristics to the blood component categories to obtain the classification result;

[0010] S5. Calculate the error according to the classification result, and adjust the blood pump flow rate through a PID controller to keep the stratified structure stable.

[0011] Further, the step S3 specifically includes the following sub-steps:

[0012] S301. Calculate the diffuse reflection light intensity component through the Lambert diffuse reflection model according to the scattering characteristics of blood components;

[0013] S302. Calculate the specular reflection light intensity component through the Phong specular reflection model according to the reflection contribution of the blood layer surface to light;

[0014] S303. Calculate the time-frequency characteristics of the reflected light intensity signal through short-time Fourier transform to obtain the time-frequency characteristic matrix.

[0015] Further, the step S4 specifically includes the following sub-steps:

[0016] S401. Integrate the extracted time-frequency characteristics, the obtained time change rate characteristics of the reflected light intensity, and the optical characteristics of the obtained reflected light characteristic data to construct the input feature vector of the neural network;

[0017] S402. Classify through the hierarchical perception radial basis function neural network;

[0018] S403. Determine the blood component categories, and the component categories at least include PRP, red blood cells, white blood cells, and plasma.

[0019] Further, step S5 specifically includes the following sub-steps:

[0020] S501. Obtain the target stratification state and the actual stratification state according to the classification result output by the hierarchical perception radial basis function neural network;

[0021] S502. Calculate the error according to the difference between the target stratification state and the actual stratification state;

[0022] S503. Dynamically adjust the proportional gain of the PID controller according to the optical characteristics of the blood component layer;

[0023] S504. Optimize the adjustment of the blood pump through the change of the blood layer position, and dynamically adjust the integral gain of the PID controller;

[0024] S505. Dynamically adjust the derivative gain of the PID controller according to the change rate of the blood stratification;

[0025] S506. Output signals through the adjusted PID controller, and adjust the flow rate of the blood pump through the output signals.

[0026] Further, the target stratification state and the actual stratification state respectively represent the position and thickness of the target stratification and the position and thickness of the layer measured according to the light reflection intensity.

[0027] Further, in step S503, the specific process of dynamically adjusting the proportional gain of the PID controller is as follows: Set the dynamic proportional gain, and combine the set dynamic proportional gain to adjust the proportional factor of the PID controller according to the real-time state of the blood stratification, where the dynamic proportional gain is expressed as:

[0028] ;

[0029] Wherein, the represents the dynamic proportional gain, the represents the original proportional gain of the PID controller, the represents the adjustment factor for controlling the change range of the proportional gain, the represents the reflection ratio of the current data, the represents the reflection ratio of the reference data, and the represents the optical characteristic difference of the blood component.

[0030] Further, in step S504, the specific process of dynamically adjusting the integral gain of the PID controller is as follows: Set the dynamic integral gain, and reduce the excessive accumulation of the integral term of the PID controller by adjusting the integral gain in real time, where the dynamic integral gain is expressed as:

[0031] ;

[0032] Among them, the represents the dynamic integral gain, the represents the original integral gain of the PID controller, the represents the adjustment factor for controlling the response of the integral gain to the error magnitude, and the represents the error at time t, that is, the difference between the target stratification state and the actual stratification state.

[0033] Furthermore, in the step S505, the specific process of dynamically adjusting the derivative gain of the PID controller is as follows: Set the derivative gain of the prediction model based on the light reflection fluctuation, and control the mutation process of the blood stratification state by predicting the change trend of the blood layer. Among them, the derivative gain of the prediction model based on the reflection fluctuation is expressed as:

[0034] ;

[0035] Among them, the represents the derivative gain of the prediction model based on the reflection fluctuation, the represents the original derivative gain of the PID controller, the represents the adjustment factor for controlling the response of the derivative gain to the change rate, and the represents the diffuse reflection light intensity / specular reflection light intensity.

[0036] Furthermore, in the step S2, the specific process of calculating the time change rate of the emitted light intensity is expressed as:

[0037] ;

[0038] Among them, the represents the time change rate of the emitted light intensity, the represents the reflected light intensity at the (N + 1)-th acquisition, and the represents the reflected light intensity at the N-th acquisition.

[0039] Furthermore, in the step S2, the specific process of calculating the second-order change rate of the reflected light intensity is expressed as:

[0040] ;

[0041] Among them, the represents the second-order change rate of the reflected light intensity, which is used to reflect the acceleration change of the light intensity.

[0042] The beneficial effects of the invention are:

[0043] The present invention combines optical detection and neural network classification techniques to achieve high-precision recognition and dynamic separation of blood components. Specifically, the present invention effectively separates different light components through the Lambert diffuse reflection model and the Phong specular reflection model, and further extracts the time-frequency characteristics of the blood layer through short-time Fourier transform to enhance the accuracy of component recognition. In addition, the present invention uses a hierarchical perception radial basis function neural network for classification, maps the optical features to specific blood component categories, and realizes the simultaneous recognition and separation of multiple components; and dynamically adjusts the blood pump flow rate through a PID controller to ensure the stability of the hierarchical structure and improve the real-time performance and controllability of the separation process. Brief Description of the Drawings

[0044] Figure 1 It is a flowchart of a method for collecting blood components provided by an embodiment of the present invention;

[0045] Figure 2 It is a structural diagram of a device for a method for collecting blood components provided by an embodiment of the present invention;

[0046] Figure 3 It is a light reflection diagram of a multi-point light source for a method for collecting blood components provided by an embodiment of the present invention;

[0047] Figure 4 It is a light reflection diagram of a bundle light source for a method for collecting blood components provided by an embodiment of the present invention. Detailed Description of the Specific Embodiment

[0048] The technical solution of the present invention will be further described in detail below with reference to the drawings, but the protection scope of the present invention is not limited to the following.

[0049] In order to make the purpose, technical solution and advantages of the present invention clearer, the present invention will be further described in detail in combination with the drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention, that is, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. Usually, the components of the embodiments of the present invention described and shown in the drawings here can be arranged and designed in various different configurations.

[0050] Therefore, the following detailed description of the embodiments of the present invention provided in the accompanying drawings is not intended to limit the scope of the claimed invention, but merely represents selected embodiments of the present invention. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative efforts fall within the scope of protection of the present invention. It should be noted that relational terms such as "first" and "second" are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any actual relationship or order between these entities or operations.

[0051] Moreover, the term "comprising", "including" or any other variation thereof is intended to cover non-exclusive inclusion, such that a process, method, article or apparatus comprising a series of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article or apparatus. Without further limitation, an element defined by the phrase "comprising one..." does not exclude the presence of additional identical elements in the process, method, article or apparatus comprising the element.

[0052] The features and performance of the present invention will be further described in detail below in conjunction with embodiments.

[0053] Among them, as Figure 1 , a blood component collection method includes the following steps:

[0054] S1. Pump whole blood into the separation chamber through a whole blood pump, and rotate at high speed by a motor, so that the whole blood in the separation chamber is stratified according to the different densities of blood components. At the same time, emit light from a light source to irradiate different blood component layers in the separation chamber, and collect the time series reflected light intensity;

[0055] S2. Calculate the time change rate of the reflected light intensity and the second-order change rate of the reflected light intensity according to the collected time series reflected light intensity;

[0056] S3. Separate different light components through the Lambert diffuse reflection model and the Phong specular reflection model to obtain the optical characteristics of the reflected light intensity; perform time-frequency feature extraction on the blood layer through short-time Fourier transform;

[0057] S4. Construct the input feature vector of the neural network, input the input vector into the hierarchical perception radial basis function neural network for classification, and map the optical characteristics to the blood component categories to obtain the classification result;

[0058] S5. Calculate the error according to the classification result, and adjust the blood pump flow rate through a PID controller to maintain the stability of the stratified structure.

[0059] Specifically, for the constructed feature vector, it contains the following data:

[0060] Reflected light intensity value: It represents the reflected light characteristics of the blood layer at the current moment.

[0061] First-order change rate of light intensity: It is used to detect the change trend and distinguish the transition layers between plasma, red blood cells and PRP.

[0062] Second-order change rate of light intensity: It helps to detect the boundary changes and more accurately identify the demarcation points between blood layers.

[0063] Diffuse reflection intensity: Calculated according to the Lambert reflection model, it is related to the scattering characteristics of red blood cells.

[0064] Specular reflection intensity: Calculated according to the Phong reflection model, the PRP layer has higher specular reflection characteristics.

[0065] Time-frequency characteristics: The time-frequency characteristics extracted by short-time Fourier transform are used to detect the dynamic light intensity changes.

[0066] Furthermore, the step S3 specifically includes the following sub-steps:

[0067] S301. According to the scattering characteristics of blood components, calculate the diffuse reflection light intensity component through the Lambert diffuse reflection model;

[0068] S302. According to the reflection contribution of the blood layer surface to light, calculate the specular reflection light intensity component through the Phong specular reflection model;

[0069] S303. Calculate the time-frequency characteristics of the reflected light intensity signal through short-time Fourier transform to obtain the time-frequency characteristic matrix.

[0070] Furthermore, the step S4 specifically includes the following sub-steps:

[0071] S401. Integrate the extracted time-frequency characteristics, the obtained time change rate characteristics of the reflected light intensity, and the optical characteristics of the obtained reflected light characteristic data to construct the neural network input feature vector;

[0072] S402. Classify through a hierarchical perception radial basis function neural network;

[0073] S403. Determine the blood component categories, and the component categories at least include PRP, red blood cells, white blood cells and plasma.

[0074] Furthermore, the step S5 specifically includes the following sub-steps:

[0075] S501. Obtain the target stratification state and the actual stratification state according to the classification result output by the hierarchical perception radial basis function neural network;

[0076] S502. Calculate the error according to the difference between the target stratification state and the actual stratification state;

[0077] S503. Dynamically adjust the proportional gain of the PID controller according to the optical characteristics of the blood component layer;

[0078] S504. Optimize the regulation of the blood pump through the change of the blood layer position, and dynamically adjust the integral gain of the PID controller;

[0079] S505. Dynamically adjust the derivative gain of the PID controller according to the change rate of the blood stratification;

[0080] S506. Output signals through the adjusted PID controller, and adjust the flow rate of the blood pump through the output signals.

[0081] Further, the radial basis function neural network is composed of an input layer, a hidden layer and an output layer; the input layer is used to receive feature vectors; the hidden layer is used to calculate the similarity between the input data and the center point using the radial basis function; the output layer is used to calculate the final classification result through weights. For the radial basis function neural network, the existing radial basis functions basically adopt ordinary Gaussian kernel functions, but this Gaussian kernel function is not stable enough in the application of the blood layer boundary region with large optical characteristic changes. Therefore, in this embodiment, the hierarchical perception radial basis function is designed by optimizing the characteristics of the light reflection intensity change, making it more sensitive to the optical characteristic changes of different blood component layers. Specifically, by analyzing the light reflection characteristics of different blood layers, it is found that:

[0082] The PRP layer has a high specular reflection component and weak diffuse reflection;

[0083] The red blood cell layer is mainly diffuse reflection and has almost no specular reflection;

[0084] The plasma layer is between the two, and the reflection characteristics are relatively balanced.

[0085] Therefore, by optimizing the RBF kernel function, making it more sensitive to the ratio of specular reflection to diffuse reflection, specifically:

[0086] ;

[0087] ;

[0088] Among them, the represents the ratio of specular reflection to diffuse reflection, and the represents, and the represents the center point of the i-th RBF kernel function, the represents the expansion scale parameter of the RBF kernel function, which is used to control the "width" of the kernel function, the represents an adjustment factor for controlling the influence degree of the light reflection ratio, the represents the feature vector of the input data in the previous step, the and respectively represent the light reflection ratios of the current data point and the center point, the makes the kernel function pay more attention to the changes in light reflection characteristics, the is responsible for basic feature matching, the and respectively represent the specular reflection light intensity and the diffuse reflection light intensity.

[0089] Specifically, for , that is, the input data to each center point distance , its calculation process is:

[0090] ;

[0091] Among them, the represents the total number of indexes, the represents the index, the represents the input feature of the j-th dimension, the represents the value of the center point of the j-th dimension;

[0092] In addition, when calculating the distance, a weight adjustment of the light reflection ratio is introduced:

[0093] ;

[0094] Among them, the represents the input data with the weight adjustment of the light reflection ratio introduced to each center point distance, the represents the input data to each center point distance, the represents the weight parameter, making the light reflection ratio play a greater role in distance calculation;

[0095] Furthermore, the expansion scale parameter is calculated through ; among them, the represents, the represents the index of the neuron, the represents the number of neurons in the hidden layer, the represents the center point of the i-th RBF kernel function, and the represents the center point of the j-th RBF kernel function;

[0096] Furthermore, the adjustment factor for controlling the influence degree of the light reflection ratio is calculated through ; through the adjustment factor, it is ensured that the differences in light reflection characteristics of different blood layers are fully utilized to improve the classification accuracy.

[0097] In summary, the final optimized RBF neural network classification is expressed as:

[0098] ;

[0099] wherein, the represents the activation value of the k-th class, which is determined by the weighted output value of the RBF neural network and is used to calculate the classification probability. The represents the weight of the k-th class corresponding to the blood component category and the i-th RBF center. The k represents the number of blood component categories, and the categories at least include PRP, red blood cells, white blood cells, and plasma.

[0100] For the traditional Gaussian kernel function, the optimized light reflection ratio-sensitive RBF kernel function enhances the perception of the optical characteristics of blood components. By introducing the ratio of specular reflection to diffuse reflection, it improves the classification accuracy; provides better discrimination ability at the blood layer junction, reduces misclassification, and improves the accuracy of component boundary recognition; improves the response ability to real-time data, can more quickly identify changes in the blood component layer, and improves the system stability. Specifically, the reflected light consists of ambient light, diffuse reflection, and specular reflection, and different blood components reflect light in different ways. Therefore, through the Lambert diffuse reflection model and the Phong specular reflection model, different light components are separated to extract the characteristics of the blood layer.

[0101] Furthermore, the target stratification state and the actual stratification state respectively represent the position and thickness of the target stratification and the position and thickness of the layer measured according to the light reflection intensity.

[0102] Furthermore, in the step S503, the specific process of dynamically adjusting the proportional gain of the PID controller is as follows: set the dynamic proportional gain, and in combination with the set dynamic proportional gain, adjust the proportional factor of the PID controller according to the real-time state of blood stratification. Among them, the dynamic proportional gain is expressed as:

[0103] ;

[0104] wherein, the represents the dynamic proportional gain, and the represents the original proportional gain of the PID controller. represents an adjustment factor for controlling the range of change of the proportional gain, and the represents the reflection ratio of the current data, and the represents the reflection ratio of the reference data, and the represents the optical property difference of the blood components.

[0105] Further, in the step S504, the specific process of dynamically adjusting the integral gain of the PID controller is as follows: setting a dynamic integral gain, and reducing the excessive accumulation of the integral term of the PID controller by adjusting the integral gain in real time, where the dynamic integral gain is expressed as:

[0106] ;

[0107] wherein the represents the dynamic integral gain, and the represents the original integral gain of the PID controller, and the represents an adjustment factor for controlling the response of the integral gain to the error magnitude, and the represents the error at time t, that is, the difference between the target stratification state and the actual stratification state.

[0108] Further, in the step S505, the specific process of dynamically adjusting the derivative gain of the PID controller is as follows: setting the derivative gain of the prediction model based on the light reflection fluctuation, and controlling the mutation process of the blood stratification state by predicting the change trend of the blood layer, where the derivative gain of the prediction model based on the reflection fluctuation is expressed as:

[0109] ;

[0110] wherein the represents the derivative gain of the prediction model based on the reflection fluctuation, and the represents the original derivative gain of the PID controller, and the represents an adjustment factor for controlling the response of the derivative gain to the change rate, and the represents the diffuse reflection light intensity / specular reflection light intensity.

[0111] Further, in the step S2, the specific process of calculating the time change rate of the emitted light intensity is expressed as:

[0112] ;

[0113] wherein the represents the time change rate of the emitted light intensity, and the represents the reflected light intensity at the (N + 1)-th acquisition, and the represents the reflected light intensity at the N-th acquisition.

[0114] Further, in step S2, the specific process of calculating the second-order change rate of the reflected light intensity is expressed as:

[0115] ;

[0116] wherein, the represents the second-order change rate of the reflected light intensity and is used to reflect the acceleration change of the light intensity.

[0117] Further, the improvement of the PID controller is the core content of this embodiment. Specifically, the light reflection intensity data of blood components are classified by an RBF neural network to obtain the stratification status and layer thickness information of different blood components (such as plasma layer, PRP, white blood cell layer, red blood cell layer, etc.). The classification results are output, including the type of each component layer and its relative position. The blood stratification status and stratification thickness information are output by the neural network, and the difference between the target layer (such as the PRP layer) and the actual layer (the layer obtained according to the light reflection intensity measurement) is calculated. The target stratification status and the actual stratification status obtained from the classification results are used to calculate the error, which represents the difference between the target stratification status and the actual stratification status and serves as the input of the PID control. The error is calculated by the difference between the target stratification position and the current stratification position. This error reflects the deviation of blood stratification and directly affects the subsequent PID control process. The error and the light reflection intensity data are used to represent the actual characteristics of the blood layer (such as the optical characteristic differences of different layers).

[0118] Further, in the blood stratification system, the light reflection intensity of blood components changes non-linearly with the change of component layers, resulting in the system's response to errors may not be completely linear; secondly, due to possible noise in the measurement process, traditional PID control may lead to unstable reactions.

[0119] For the above reasons, this embodiment improves the traditional PID control:

[0120] PID adjustment based on light reflection intensity: A dynamic adjustment proportional factor is introduced for the reflected light intensity differences of different blood layers, enabling it to be adjusted in real time according to different component layers to cope with the changes in the light reflection characteristics of different layers.

[0121] Optimization of the integral term: To reduce the long-term deviation caused by the change of blood layers, the integral term is optimized to ensure that the cumulative effect of the error will not cause over-regulation of the system.

[0122] Adjustment of the differential term: A light reflection fluctuation prediction model is introduced, and the differential term is used to predict the change trend of blood layers to improve the real-time performance and accuracy of the response and avoid over-control reactions caused by unstable fluctuations.

[0123] Specifically, due to the large difference in the light reflection intensity of the blood component layers, by setting a dynamic proportional gain, the PID control adjusts the proportional factor according to the real-time state of blood stratification, and through the dynamic proportional gain, the response ability to the difference in light reflection intensity is improved, achieving more precise stratification control.

[0124] Furthermore, the integral term is used to eliminate long-term stable errors, but when set too high, it will cause over-regulation. By setting a dynamic integral gain, real-time adjustment is made to reduce the excessive accumulation of the integral term. The gain of the integral term will be dynamically adjusted according to the change of the error, avoiding the integral windstorm in traditional PID control.

[0125] Furthermore, the derivative term is used to predict the trend of error change and reduce system over-regulation. In order to make the derivative term better respond to the sudden change of blood stratification state, by setting the derivative gain of the prediction model based on light reflection fluctuation, the derivative term can sensitively respond to the change of the blood layer, especially during the sudden change of blood stratification state, improving the stability of system response.

[0126] Combining the optimized proportional, integral, and derivative terms above, the final PID control algorithm is specifically as follows:

[0127] ;

[0128] wherein, the represents the output signal of the PID control algorithm, that is, the blood pump flow rate adjustment control signal.

[0129] Furthermore, as a preferred implementation of the above embodiment, a collection device for applying a method of collecting blood components is proposed, including a motor, a blood separation chamber, an optical detector, a controller, and a blood pipeline; wherein:

[0130] The blood separation chamber stratifies blood components through centrifugal action;

[0131] The optical detector irradiates different blood components, such as plasma, platelets, white blood cells, and red blood cells, with a light source, and the light receiver judges the blood components and their concentrations by receiving the change in the light intensity of the reflected light;

[0132] The controller includes a pump and a valve for controlling fluid flow and an information processor;

[0133] The information processor includes a programmable processor, wherein, the processor is embedded with a PID algorithm, and this algorithm is based on the PID control algorithm with optimized proportional, integral, and derivative terms described in the above embodiment;

[0134] The pipeline contains blood and blood components and is connected to the controller and the blood separation chamber.

[0135] Further, as a preferred implementation of the above embodiments, a device for collecting blood components is proposed, including a motor, a blood separation chamber, an optical detector, an information processor, a controller, and a blood pipeline. As shown in Figure 2, the motor provides the power for blood centrifugal separation. The blood separation chamber contains blood and rotates at a high speed with the motor to layer the blood in the separation chamber according to blood components. The optical detector includes a light emission source and a light receiver. The light emission source is a point, multiple points, or a fiber bundle, and the light receiver is a single point, multiple points, or a bundle shape and corresponds to the emitted light. For example Figure 3 and 4 , the light emission source emits light, directly irradiates the blood in the blood separation chamber and reflects the light, and the reflected light is received by the light receiver. The information processor calculates according to the light intensity of the reflected light and issues an instruction to the controller. The controller receives the instruction issued by the information processor and controls the pump speeds of multiple pumps.

[0136] The above is only the preferred implementation manner of the present invention. It should be understood that the present invention is not limited to the form disclosed herein, should not be regarded as excluding other embodiments, but can be used in various other combinations, modifications, and environments, and can be changed within the scope of the concept described herein through the above teachings or the technology or knowledge in related fields. And the changes and modifications made by those skilled in the art without departing from the spirit and scope of the present invention should be within the protection scope of the appended claims of the present invention.

Claims

1. A method for collecting blood components, characterized in that: The following steps are involved: S1. Pumping whole blood into the separation chamber through a whole blood pump and rotating the motor at high speed, so that the whole blood in the separation chamber is stratified according to the density of the blood components, and at the same time emitting light from a light source to illuminate the different blood component layers in the separation chamber, and collecting the reflected light intensity in a time series; S2. calculating the time change rate of the reflected light intensity and the second-order change rate of the reflected light intensity according to the collected time series reflected light intensity; S3. Separate different light components through Lambert diffuse reflection model and Phong specular reflection model to obtain the optical characteristics of the intensity of reflected light; extract the time-frequency characteristics of the blood layer through short-time Fourier transform; S4. constructing an input feature vector of the neural network, inputting the input vector into a hierarchical perceptual radial basis function neural network for classification, and mapping the optical features to blood component categories to obtain classification results; S5. Calculate the error based on the classification result and adjust the blood pump flow rate through the PID controller to keep the layered structure stable.

2. A blood component collection method according to claim 1, characterized in that: The step S3 specifically includes the following sub-steps: S301. Calculating the diffuse reflection light intensity component by Lambert diffuse reflection model according to the scattering characteristics of blood components; S302. Calculating the specular reflection light intensity component by the Phong specular reflection model according to the reflection contribution of the blood layer surface to the light; S303. Calculate the time-frequency characteristics of the reflected light intensity signal by short-time Fourier transform to obtain a time-frequency feature matrix.

3. A blood component collection method according to claim 1, characterized in that: The step S4 specifically includes the following sub-steps: S401. Integrate the extracted time-frequency features, the acquired time-varying rate features of the reflected light intensity, and the acquired optical features of the reflected light feature data to construct a neural network input feature vector; S402. Classification by layered perceptual radial basis function neural network; S403. Determine the blood component categories, which include at least PRP, red blood cells, white blood cells and plasma.

4. A blood component collection method according to claim 1, characterized in that: The step S5 specifically includes the following sub-steps: S501. According to the classification result output by the layered-aware radial basis function neural network, the target layered state and the actual layered state are obtained; S502. Calculating the error based on the difference between the target layered state and the actual layered state; S503. Dynamically adjust the proportional gain of the PID controller according to the optical properties of the blood component layer; S504. Optimizing the regulation of the blood pump by changing the position of the blood layer, and dynamically adjusting the integral gain of the PID controller; S505. Dynamically adjust the differential gain of the PID controller according to the rate of change of blood stratification; S506. The adjusted PID controller is used to output a signal, and the flow rate of the blood pump is adjusted by the output signal.

5. A blood component collection method according to claim 4, characterized in that: The target delamination state and the actual delamination state respectively represent the position and thickness of the target delamination and the position and thickness of the layer obtained according to the light reflection intensity measurement.

6. A blood component collection method as claimed in claim 4, characterized in that: In step S503, the specific process of dynamically adjusting the proportional gain of the PID controller is: setting the dynamic proportional gain, and adjusting the proportional factor of the PID controller according to the real-time state of blood stratification in combination with the set dynamic proportional gain, wherein the dynamic proportional gain is expressed as: ； Among them, the represents the dynamic proportional gain, the Represents the original proportional gain of the PID controller, Represents the adjustment factor used to control the variation range of the proportional gain. Indicates the reflectance value of the current data. represents the reflectance value of the reference data, Indicates the differences in optical properties of blood components.

7. A blood component collection method according to claim 4, characterized in that: In step S504, the specific process of dynamically adjusting the integral gain of the PID controller is: setting a dynamic integral gain, and reducing excessive accumulation of the integral term of the PID controller by adjusting the integral gain in real time, wherein the dynamic integral gain is expressed as: ； Among them, the Represents the dynamic integral gain, the Represents the original integral gain of the PID controller. represents the adjustment factor used to control the response of the integral gain to the error size, It represents the error at time t, that is, the difference between the target stratification state and the actual stratification state.

8. A blood component collection method according to claim 4, characterized in that: In step S505, the specific process of dynamically adjusting the differential gain of the PID controller is as follows: setting the differential gain of the prediction model based on light reflection fluctuation, and controlling the mutation process of the blood stratification state by predicting the change trend of the blood layer, wherein the differential gain of the prediction model based on reflection fluctuation is expressed as: ； Among them, the represents the differential gain of the prediction model based on reflection fluctuation, Represents the original differential gain of the PID controller, represents the adjustment factor used to control the response of the derivative gain to the rate of change, Indicates diffuse light intensity / specular light intensity.

9. A blood component collection method according to claim 1, characterized in that: In step S2, the specific process of calculating the time change rate of the emitted light intensity is expressed as follows: ； Among them, the represents the time rate of change of the emitted light intensity, Indicates the reflected light intensity during the N+1th acquisition. Indicates the intensity of reflected light at the Nth acquisition.

10. A blood component collection method according to claim 9, characterized in that: In step S2, the specific process of calculating the second-order change rate of the reflected light intensity is expressed as follows: ； Among them, the It represents the second-order rate of change of the reflected light intensity and is used to reflect the acceleration change of light intensity.

Images