Oxygen consumption adjusting respirator based on physical sign feature machine learning and adjusting method of oxygen consumption adjusting respirator

Abstract

The invention relates to the technical field of mine rescue equipment, in particular to an oxygen consumption adjusting respirator based on physical sign feature machine learning and an adjusting method thereof, and the adjusting method comprises the following steps: data acquisition: acquiring physical sign data and oxygen pressure data of rescue workers in rescue work; establishing a machine learning oxygen consumption prediction model, performing machine learning analysis on the imported physical sign data and oxygen pressure data by using an algorithm model, and establishing the machine learning oxygen consumption prediction model; oxygen consumption intelligent adjustment is conducted, real-time physical sign data and oxygen pressure data of rescue workers in actual rescue work are obtained, the optimal oxygen consumption is calculated based on a machine learning oxygen consumption prediction model, and a microcontroller is used for controlling an intelligent adjusting valve to reasonably adjust the oxygen supply amount. The oxygen consumption intelligent adjustment of the respirator can be realized according to the individual physiological sign oxygen consumption demand of a user, so that the mine emergency rescue personnel are in the optimal breathing state in the rescue work, and the rescue efficiency is improved.

Description

Technical Field

[0001] The present invention relates to a technical field, specifically an oxygen consumption regulating respirator based on machine learning of vital sign characteristics and a regulating method thereof. Background Art

[0002] Mine emergency rescue is the last line of defense for the safety of miners' lives. Successful emergency rescue operations can save a large number of lives and property safety. In recent years, with the continuous improvement of mine emergency rescue equipment and technology, the effectiveness of mine accident rescue has made significant progress. As of 2023, the number of personnel in my country's mine emergency rescue professional teams has exceeded 30,000, and there are countless part-time rescue team members. In the process of mine disaster relief, mine oxygen respirators are the most important individual protective equipment for underground workers (especially rescue team members), and the performance of the respirator directly determines the efficiency of mine emergency rescue. However, the current intelligence level of foreign mine oxygen respirators is low. Due to the limitation of volume and weight, the continuous oxygen supply is only about 4 hours. Under urgent rescue working conditions, the oxygen consumption of oxygen respirators is difficult to adapt to the needs of users and effectively adjust, causing rescuers to have difficulty breathing and increase the rescue burden, which seriously restricts the safe and efficient implementation of rescue operations in disaster areas.

[0003] There are many types of respirators at present, including isolation type, positive pressure oxygen respirators, etc. Each type of respirator has its own characteristics, and mine personnel choose to use it according to their own needs. These respirator adjustment methods are usually: adjust the oxygen supply by detecting the oxygen status in the breathing circuit. When the wearer's workload is small and the respirator's oxygen supply is greater than the oxygen consumption, the oxygen supply is reduced; when the wearer's workload is large and the respirator's oxygen supply is less than the oxygen consumption at this time, the oxygen supply is increased. For example, the prior art "Oxygen Respirator and Oxygen Supply Method for Oxygen Respirator" (Publication No.: CN114681831A) detects the oxygen concentration, gas pressure and gas flow parameters in the breathing circuit of the respirator by using a detector, and triggers different adjustment signals according to whether the detected parameters reach the preset range value, thereby controlling the opening of the respirator solenoid valve to adjust the oxygen supply, but the prior art still has the following technical problems:

[0004] The oxygen supply is not precisely regulated. In actual rescue work, the rescue environment is extremely complex, and the environment of each rescue work is different. At the same time, different rescuers are different in physical and psychological qualities. This will lead to the fact that the existing technology of feedback regulation by detecting whether the various parameters in the respirator have reached the preset value is not accurate. For example, when the oxygen supply signal of the respirator is triggered, due to the influence of environmental factors and individual differences of rescuers, different rescuers have different actual demands for oxygen content, some with more or some with less. If the oxygen supply signal is triggered uniformly according to the same preset value standard, some rescuers will suffer from hypoxia due to insufficient oxygen supply, threatening the life safety of the rescuers; or because of excessive oxygen supply, the oxygen accumulates in the airbag for too long and is discharged to the external environment, causing oxygen waste, thereby shortening the use time of the oxygen cylinder and reducing the rescue efficiency. Summary of the invention

[0005] The present invention provides an oxygen consumption regulating respirator based on machine learning of vital sign characteristics and a regulating method thereof, which can solve the problems in the prior art of inaccurate regulation of the oxygen supply of the respirator, which easily leads to insufficient oxygen supply and threatens the lives of rescuers, and excessive oxygen supply leads to oxygen waste, shortening the use time of oxygen cylinders and reducing rescue efficiency.

[0006] The present application provides the following technical solution, a method for regulating an oxygen consumption regulating respirator based on machine learning of vital sign characteristics; comprising the following steps:

[0007] Data collection: use smart watches to collect the vital signs data of rescuers during rescue work, and use oxygen pressure sensors to collect oxygen pressure data in the oxygen supply pipeline of the respirator. The vital signs data and oxygen pressure data are imported into the smart watch to calculate and save the oxygen consumption;

[0008] Establish a machine learning oxygen consumption prediction model, and import the collected vital sign data and oxygen pressure data into the model for machine learning analysis, obtain and save the oxygen consumption data under different vital sign data and different oxygen pressure data;

[0009] Intelligent oxygen consumption regulation obtains real-time vital signs data and oxygen pressure data of rescuers during actual rescue work, calculates the optimal oxygen consumption based on the machine learning oxygen consumption prediction model, and uses Bluetooth to transmit the optimal oxygen consumption signal to the microcontroller on the respirator. The microcontroller controls the intelligent regulating valve to reasonably adjust the oxygen supply.

[0010] Beneficial effects:

[0011] 1. Improve the accuracy of oxygen supply. When conventional breathing devices supply oxygen, the oxygen tank usually automatically replenishes oxygen to the airbag to keep the airbag within a certain safe pressure range. However, the amount of oxygen consumed by rescuers in different environments is not fixed. For example, in a temporary emergency, the rescuers need a large amount of oxygen. However, since the airbag uses pressure to adjust the oxygen supply, although the airbag pressure is within a safe range, the actual oxygen demand of the rescuers is greater. The amount of oxygen stored in the airbag under normal pressure can no longer meet the needs of the rescuers. If the oxygen supply is not increased at this time, the rescuers will suffer from hypoxia, posing a threat to their lives. Therefore, it is inaccurate to judge the oxygen demand solely by the airbag pressure, which can easily cause the problem of insufficient oxygen supply in an emergency. The inventor has improved many solutions, including controlling the oxygen delivery flow rate and trying to adjust the preset pressure range in the airbag, but none of them can achieve good results. At the same time, due to the differences in individual characteristics of rescuers, the actual oxygen demand required by different rescuers under different environmental conditions is also different, which further increases the difficulty in controlling the oxygen supply. After many failed attempts, The inventor found that in the early stage, he had been considering how to improve the accuracy of oxygen supply from the oxygen supply side, but the effect achieved was not good. Therefore, the inventor changed the research and development ideas, starting from the demand side, considering the oxygen demand corresponding to the rescue personnel's vital signs data in different environments to reversely adjust the oxygen supply. Therefore, the inventor re-formulated the technical plan, established a machine learning oxygen consumption prediction model, calculated and analyzed the multiple vital signs data of the rescue personnel during each rescue and training work and the oxygen pressure data during oxygen supply, and obtained the oxygen consumption data under different states, so that the optimal oxygen consumption in the corresponding working environment can be adapted for the rescue personnel, and then oxygen delivery can be accurately provided according to the optimal oxygen consumption. Compared with the existing technology, this solution can provide the most appropriate oxygen supply according to the actual physical condition of the rescue personnel, so that the rescue personnel are always in the most comfortable breathing oxygen consumption state, so as to improve the efficiency of the rescue; at the same time, it can avoid dangerous situations such as hypoxia of the rescue personnel caused by insufficient oxygen delivery, ensure the life safety of the rescue personnel, and avoid excessive oxygen delivery and oxygen waste, so as to extend the use time of oxygen cylinders and improve the rescue efficiency.

[0012] 2. Improve the reliability of oxygen consumption prediction. Since the physical and psychological qualities of rescuers will continue to change over time, the machine learning oxygen consumption prediction model can obtain the rescuers' vital signs data and oxygen pressure data during each work or training, and continuously update and optimize the data, so that the model can always change with the changes in the physical and mental state of the rescuers, adapt to the best oxygen consumption for the rescuers in the current period, improve the reliability of the model's oxygen consumption prediction, and ensure the oxygen supply effect.

[0013] Furthermore, the establishment of the machine learning oxygen consumption prediction model includes the following steps: data processing, cleaning the collected vital sign data and oxygen pressure data; data modeling, modeling with the vital sign data monitored by the smart watch as feature variables; data set division, dividing the collected vital sign data and oxygen pressure data into a training set and a test set, the training set is used to train the machine learning oxygen consumption prediction model, and the test set is used to evaluate the performance of the trained model prediction; model parameter adjustment, adjusting the various parameters in the machine learning oxygen consumption prediction model to a reasonable range; model correction, continuously introducing new vital sign data and oxygen pressure data to optimize the model.

[0014] Beneficial effects: Data processing can improve data quality and avoid the impact of abnormal data on model prediction results; modeling with vital sign data as feature variables can enable the model to more accurately predict oxygen consumption based on individual differences among rescuers; reasonably dividing the data set into training sets and test sets can ensure that the model has good generalization ability and improve the model's adaptability when processing new data; at the same time, continuously introducing new data can enable the model to continuously learn and correct, ensuring that the model can adapt to new data changes and improve the reliability of the model.

[0015] Furthermore, in the data processing step of establishing the machine learning oxygen consumption prediction model, the missing values ​​and outliers in the collected vital sign data and oxygen pressure data are cleaned and deduplicated.

[0016] Beneficial effects: By processing missing values ​​and outliers, data quality is improved, the deviation of model prediction results caused by abnormal data is eliminated, and the effectiveness and reliability of model training are ensured.

[0017] Furthermore, in the data modeling step of establishing a machine learning oxygen consumption prediction model, the vital sign data include blood oxygen, heart rate, body temperature and exercise status.

[0018] Beneficial effects: Variables such as blood oxygen, heart rate, body temperature, and exercise status are closely related to oxygen consumption. Using these as characteristic variables for modeling can improve the accuracy of the oxygen consumption results predicted during model training, which is beneficial to ensuring the oxygen supply effect for rescue personnel.

[0019] Furthermore, the machine learning oxygen consumption prediction model includes multiple decision trees, each decision tree includes multiple sub-items, and different vital sign data are randomly extracted into each sub-item so that each decision tree can match a corresponding oxygen consumption.

[0020] Beneficial effects: Through the machine learning oxygen consumption prediction model, different blood oxygen, heart rate, body temperature and exercise status are randomly extracted into multiple sub-items of each decision tree, so that each decision tree matches the corresponding oxygen consumption under the group of vital signs. After multiple machine learning, it can accurately predict the most appropriate oxygen consumption to supply when the same or similar vital signs data is detected next time. The more data is learned, the more accurate the predicted oxygen consumption results will be. It not only greatly improves the accuracy of oxygen supply, but also can continuously optimize and improve the model with the user's real-time vital signs data, and has wide applicability.

[0021] An oxygen consumption regulating respirator based on machine learning of vital signs and characteristics can use the above-mentioned regulation method of the oxygen consumption regulating respirator based on machine learning of vital signs and characteristics, including a smart watch, an oxygen cylinder, an air bag and an oxygen supply system connected between the oxygen cylinder and the air bag, the oxygen supply system includes an oxygen supply pipeline, a cooler, an inhalation hose, a connecting pipe, an exhalation hose and an absorption tank connected in sequence, the oxygen supply pipeline is connected to the oxygen cylinder at one end away from the cooler, the cooler is also connected to the air bag, the absorption tank is connected to the air bag at one end away from the exhalation hose, and the connecting pipe is used to connect to the breathing mask.

[0022] Beneficial effects: When replenishing oxygen, the oxygen in the oxygen cylinder first flows into the cooler through the oxygen supply pipeline and then into the airbag. The cooler can reduce the temperature of the oxygen, making the breathable gas cooler in the high temperature outdoor environment, increasing the breathing comfort of the rescuers; the inhalation hose facilitates the delivery of the gas in the airbag into the connecting pipe and then into the breathing mask during inhalation, and the absorption tank can absorb the carbon dioxide exhaled from the exhalation hose, maintain the stable ratio of air components in the airbag, and prevent carbon dioxide from entering the airbag and causing carbon dioxide poisoning to the rescuers during inhalation.

[0023] Furthermore, the oxygen supply pipeline is also provided with a pressure reducer and an intelligent regulating valve, and the intelligent regulating valve is located between the pressure reducer and the cooler.

[0024] Beneficial effects: Since the oxygen cylinder is in a high-pressure state, the pressure reducer is used to reduce the pressure of the oxygen flowing out of the oxygen cylinder and maintain it at a low pressure level suitable for rescuers to breathe, thereby improving breathing comfort. It can also prevent the oxygen supply pipeline from bursting due to excessive pressure during oxygen delivery, and ensure that oxygen is stably delivered to the airbag.

[0025] Furthermore, the intelligent regulating valve includes a valve body, a power supply fixed in the valve body, a motor, a microcontroller and a driving mechanism, the microcontroller is electrically connected to the motor, and the driving mechanism includes a shell fixed in the valve body, a base fixed under the shell, a positioning frame fixed in the shell, a rack slidably connected to the positioning frame, a gear meshing with the rack, a connecting column fixed at the lower end of the rack, a rubber pad fixed in the middle of the shell, a connecting rod fixed to the bottom of the rubber pad, and a compression spring sleeved on the rod of the connecting rod, the bottom of the connecting column is fixed to the upper surface of the rubber pad; a blocking hole is vertically provided in the base, and the two ends of the blocking hole are respectively provided with connected air inlet and air outlet, and the bottom end of the connecting rod can be inserted from the shell downward into the blocking hole to block the connection between the air inlet and the air outlet.

[0026] Beneficial effect: After receiving the oxygen supply signal from the smart watch, the microcontroller can control the micromotor to rotate and drive the rack to move vertically up and down, thereby driving the connecting rod to move up and down in the shell, and then controlling the bottom end of the connecting rod to be inserted into the blocking hole, changing the size of the channel between the air inlet and the air outlet, so as to accurately adjust the oxygen supply.

[0027] Furthermore, the oxygen pressure sensor is arranged on the air outlet, and the oxygen pressure sensor is electrically connected to the microcontroller.

[0028] Beneficial effects: By setting an oxygen pressure sensor at the air outlet, the pressure of oxygen during delivery can be detected, and the received oxygen pressure signal can be transmitted to the machine learning oxygen consumption prediction model in the smart watch through the microcontroller, so as to optimize the parameters of the prediction model. At the same time, it can also ensure that in actual rescue work, the pressure of oxygen supply is monitored in real time to ensure the accuracy of oxygen supply.

[0029] Furthermore, the shell includes an upper shell and a lower shell fixed together, a circle of annular protrusions are provided at the internal connection between the upper shell and the lower shell, and the outer ring of the rubber pad is fixed on the annular protrusion and the gap is sealed with glue.

[0030] Beneficial effect: The rubber pad is used to seal the space in the lower shell to prevent the oxygen in the base from leaking from the lower shell, thus avoiding waste during oxygen delivery. BRIEF DESCRIPTION OF THE DRAWINGS

[0031] Figure 1 It is the front view of the present invention.

[0032] Figure 2 for Figure 1 A magnified view of the smart control valve.

[0033] Figure 3 for Figure 2 Left side view of the middle shell.

[0034] Figure 4 It is a logic block diagram of the present invention. DETAILED DESCRIPTION

[0035] The following is further described in detail through specific implementation methods:

[0036] The marks in the drawings of the specification include: power supply 1, motor 2, microcontroller 3, fixing block 4, sleeve 5, rotating shaft 6, positioning frame 7, slide groove 71, upper shell 8, annular protrusion 81, gear 9, rack 10, connecting column 11, lower shell 12, rubber pad 13, connecting rod 14, compression spring 15, disc 16, air inlet 17, base 18, blocking hole 19, oxygen pressure sensor 20, air outlet 21, oxygen cylinder 22, pressure reducer 23, oxygen supply pipe 24, intelligent regulating valve 25, inhalation hose 26, connecting pipe 27, cooler 28, exhalation hose 29, absorption tank 30, air bag 31, manual oxygenation pipe 32, oxygen supply pipeline 33, shell 34.

[0037] Embodiment 1

[0038] like Figures 1 to 3 As shown, an oxygen consumption regulating respirator based on machine learning of vital sign features includes a smart watch, an oxygen cylinder 22, an air bag 31 and an oxygen supply system connected between the oxygen cylinder 22 and the air bag 31, the oxygen supply system includes an oxygen supply pipeline 33, a cooler 28, an inhalation hose 26, a connecting pipe 27, an exhalation hose 29 and an absorption tank 30 connected in sequence, the end of the oxygen supply pipeline 33 away from the cooler 28 is connected to the oxygen cylinder 22, the cooler 28 is also connected to the air bag 31, the end of the absorption tank 30 away from the exhalation hose 29 is connected to the air bag 31, a manual oxygenation tube 32 is connected between the oxygen supply pipeline 33 and the air bag 31, and a manual valve is provided on the manual oxygenation tube 32, so that oxygen can be manually added to the air bag 31. The connecting tube 27 is used to communicate with the breathing mask. Both ends of the connecting tube 27 are provided with one-way valves, so that oxygen can enter the mask through the inhalation hose 26 for the user to breathe, and the gas exhaled by the user will only enter the exhalation hose 29 through the connecting tube 27 and will not flow back into the inhalation hose 26 ( Figure 1 The breathing mask has been omitted. In addition, this respirator will be put into a backpack when in use. Figure 1 has been omitted).

[0039] like Figure 1 As shown, the oxygen supply pipeline 33 is also provided with a pressure reducer 23 and an intelligent regulating valve 25, and the intelligent regulating valve 25 is located between the pressure reducer 23 and the cooler 28. Figure 2 and Figure 3As shown, the intelligent regulating valve 25 includes a valve body (the valve body is omitted in the figure), a power supply 1 fixed in the valve body, a motor 2, a microcontroller 3 and a driving mechanism, the power supply 1 is a rechargeable power source, the microcontroller 3 is a PLC, the microcontroller 3 is electrically connected to the motor 2, and the driving mechanism includes a shell 34 fixed in the valve body, a base 18 fixed below the shell 34, a positioning frame 7 fixed in the shell 34, a rack 10 slidably connected to the positioning frame 7, a gear 9 meshing with the rack 10, a connecting column 11 fixed at the lower end of the rack 10, and a connecting column 11 fixed at the shell 3 4, a rubber pad 13 in the middle, a connecting rod 14 fixed to the bottom of the rubber pad 13, and a compression spring 15 sleeved on the rod of the connecting rod 14; the housing 34 includes an upper shell 8 and a lower shell 12, the rotating shaft 6 of the motor 2 horizontally extends into the upper shell 8, the gear 9 is key-connected to the end of the rotating shaft 6 of the motor 2, the rod of the rotating shaft 6 is sleeved with a shaft sleeve 5, the rotating shaft 6 can rotate in the shaft sleeve 5, and the upper and lower parts of the shaft sleeve 5 are provided with fixing blocks 4, the fixing blocks 4 are fixed on the wall surface inside the upper shell 8, and the fixing blocks 4 are used to clamp and fix the shaft sleeve 5, so as to support the rotation of the rotating shaft 6.

[0040] like Figure 2 As shown, the positioning frame 7 is hollow, which can reduce the overall weight. A vertical slide groove 71 is provided on the positioning frame 7, and the rack 10 is slidably connected in the slide groove 71; the bottom of the connecting column 11 is fixed to the center of the upper surface of the rubber pad 13, the rubber pad 13 is circular, and a circle of annular protrusions 81 are provided at the internal connection between the upper shell 8 and the lower shell 12. The outer circle of the rubber pad 13 is fixed to the annular protrusion 81 with screws, and the gap between the rubber pad 13 and the annular protrusion 81 is sealed with glue to achieve gas sealing in the lower shell 12; the bottom of the connecting column 11 is fixed to the center of the upper surface of the rubber pad 13. When the rack 10 moves downward, the rubber pad 13 can be tightened downward, so that the rubber pad 13 presents a cone shape (such as Figure 2 and Figure 3 ), and when the connecting column 11 does not tighten the rubber pad 13 downward, the rubber pad 13 is in a relaxed state with an arc rather than a tight state, so that when the rubber pad 13 is subjected to the tightening force of the connecting column 11, deformation and fatigue damage can be reduced, thereby extending the service life of the rubber pad 13.

[0041] A gasket is fixed on the top of the connecting rod 14, and the upper surface of the gasket is fixed to the center of the lower surface of the rubber pad 13. When the rack 10 moves up and down, it can drive the connecting rod 14 to move up and down at the same time. Figure 2 As shown, a disc 16 is fixed to the bottom of the lower shell 12, and the bottom end of the compression spring 15 is located above the disc 16. A circular hole is provided at the center of the disc 16, and the bottom end of the connecting rod 14 can be inserted downward into the circular hole.

[0042] like Figure 2As shown, a blocking hole 19 is vertically provided in the base 18, and the two ends of the blocking hole 19 are horizontally connected to the air inlet 17 and the air outlet 21, respectively. The bottom end of the connecting rod 14 can be inserted downward from the housing 34 into the blocking hole 19 to block the connection between the air inlet 17 and the air outlet 21. An oxygen pressure sensor 20 is horizontally led out from the air outlet 21 of the base 18, and the oxygen pressure sensor 20 is electrically connected to the microcontroller 3.

[0043] If the motor 2 fails and the bottom end of the connecting rod 14 blocks the blocking hole 19, the elastic force of the spring can push the connecting rod 14 upward to prevent the motor 2 from failing to work due to an accident, and the oxygen cylinder 22 can also supply oxygen.

[0044] The method for adjusting the oxygen consumption regulating respirator based on machine learning of vital signs features uses the above oxygen consumption regulating respirator based on machine learning of vital signs features to adjust the oxygen supply; the method comprises the following steps:

[0045] 1. Data collection: using a smart watch to collect the vital sign data of the rescuer during the rescue work, the vital sign data includes blood oxygen, heart rate, body temperature and movement status, and using an oxygen pressure sensor 20 to collect the oxygen pressure data in the respirator oxygen supply pipeline 33 in real time. The vital sign data is directly measured by the smart watch, and the oxygen pressure data is measured by the oxygen pressure sensor 20 and then transmitted to the smart watch by the microcontroller 3 using Bluetooth to calculate the oxygen consumption, and the vital sign data, oxygen pressure data and oxygen consumption data are saved to form a data set; in this embodiment, the oxygen consumption calculation step is specifically as follows: when the rescuer is in a certain working state, provide the rescuer with oxygen at different pressures, and let the rescuer score the oxygen supply at different pressures according to their own state, so as to collect the most suitable oxygen consumption in this working state;

[0046] 2. Establish a machine learning oxygen consumption prediction model, use the random forest algorithm model to perform machine learning analysis on the vital sign data and oxygen pressure data in the smart watch, and establish a machine learning oxygen consumption prediction model to obtain and save the oxygen consumption data under different vital sign data and different oxygen pressure data; Establishing a machine learning oxygen consumption prediction model includes the following steps:

[0047] 1. Data processing: cleaning and deduplication of missing values ​​and outliers in the collected vital sign data and oxygen pressure data; by processing missing values ​​and outliers, the data quality can be improved, the deviation of the model prediction results caused by abnormal data can be eliminated, and the effectiveness and reliability of model training can be ensured.

[0048] 2. Data modeling: use the vital sign data monitored by the smart watch as the feature variable for modeling; use the random forest algorithm model, that is, different blood oxygen, heart rate, body temperature and exercise status as feature variables, and there is a corresponding oxygen consumption under each state, so as to establish a basic prediction model.

[0049] 3. Data set division: the collected vital sign data and oxygen pressure data are divided into training set and test set. The training set is used to input into the basic model of machine learning oxygen consumption prediction for training, and the test set is used to evaluate the prediction performance of the basic model after training. Reasonable division of the data set into training set and test set can ensure that the model has good generalization ability and improve the adaptability of the model when processing new data. The ratio of training set to test set can be selected as 7:3 or 6:4. The preferred ratio in this embodiment is 6:4. Using a 60% training set ratio can reduce the risk of overfitting of the model, because there is more data to verify the generalization ability of the model. The more training, the better the accuracy of the prediction results. The 40% test set provides enough data to verify the performance of the model, which helps to more accurately evaluate the performance of the model on unseen data.

[0050] 4. Model parameter adjustment: According to the prediction results of the basic model, the various parameters in the machine learning oxygen consumption prediction model are adjusted to a reasonable range, so that the predicted results are closer to the data results in the test set. In this embodiment, for the same or similar physical signs, the prediction results of the training set are compared with the results in the test set, so as to dynamically adjust the number of groups, maximum depth, and minimum number of node samples of the decision tree to make the prediction results closer to the test set.

[0051] 5. Model correction: continuously introduce new vital sign data and oxygen pressure data to optimize the model. The introduction of new data can enable the model to continuously learn and correct, ensure that the model can adapt to new data changes, and improve the reliability of the model. In this embodiment, the model is continuously optimized and the reliability of the model is improved by regularly collecting new vital sign data and oxygen pressure data and performing outlier deduplication processing on the new data.

[0052] 3. Intelligent regulation of oxygen consumption. After the model training is completed, it will enter actual use. By obtaining the real-time vital signs data and oxygen pressure data of the rescue personnel in the actual rescue work, the optimal oxygen consumption is calculated based on the machine learning oxygen consumption prediction model. The smart watch then uses Bluetooth to transmit the optimal oxygen consumption signal to the microcontroller 3 on the respirator, and the microcontroller 3 controls the intelligent regulating valve 25 to reasonably adjust the oxygen supply. The regulation principle of the intelligent regulating valve 25 in this solution is as follows:

[0053] like Figure 2As shown, after receiving the optimal oxygen consumption signal predicted by the smart watch, the microcontroller 3 controls the rotating shaft 6 of the motor 2 to rotate, the gear 9 drives the rack 10 to move upward, and the connecting column 11 drives the rubber pad 13 upward and makes the bottom end of the connecting rod 14 under the rubber pad 13 move up from the blocking hole 19, so that the air inlet 17 and the air outlet 21 are connected, and the oxygen can come out of the oxygen cylinder 22 and enter the pressure reducer 23 to be transported after decompression. At the same time, the height of the connecting rod 14 moving up and down in the blocking hole 19 can change the channel space volume of the gas flow, thereby realizing the precise adjustment of the oxygen supply.

[0054] After passing through the intelligent regulating valve 25, the oxygen enters the cooler 28 for cooling. Part of it is stored in the air bag 31, and part of it enters the inhalation hose 26 and flows into the connecting pipe 27 and then enters the breathing mask for the user to breathe. The gas exhaled by the user enters the absorption tank 30 through the exhalation hose 29. The absorption tank 30 absorbs the carbon dioxide and then sends the gas into the air bag 31, thus completing the breathing cycle.

[0055] The above are only embodiments of the present invention. The invention is not limited to the fields involved in this implementation case. The common knowledge such as the known specific structure and characteristics in the scheme is not described in detail here. It should be pointed out that for those skilled in the art, several deformations and improvements can be made without departing from the structure of the present invention, which should also be regarded as the protection scope of the present invention, and these will not affect the effect of the implementation of the present invention and the practicality of the patent. The scope of protection required by this application shall be based on the content of its claims, and the specific implementation methods and other records in the specification can be used to interpret the content of the claims.

Claims

1. A method for regulating an oxygen consumption regulating respirator based on machine learning of vital signs, characterized in that: The steps include: Data collection: collect the vital signs data of rescuers during rescue work, and use oxygen pressure sensors to collect oxygen pressure data in the oxygen supply pipeline of the respirator. The vital signs data and oxygen pressure data are imported into the smart watch to calculate and save oxygen consumption; Establish a machine learning oxygen consumption prediction model, and import the collected vital sign data and oxygen pressure data into the model for machine learning analysis, obtain and save the oxygen consumption data under different vital sign data and different oxygen pressure data; Intelligent oxygen consumption regulation obtains real-time vital signs data and oxygen pressure data of rescuers during actual rescue work, calculates the optimal oxygen consumption based on the machine learning oxygen consumption prediction model, and uses Bluetooth to transmit the optimal oxygen consumption signal to the microcontroller on the respirator. The microcontroller controls the intelligent regulating valve to reasonably adjust the oxygen supply.

2. The method for adjusting an oxygen consumption regulating respirator based on machine learning of vital signs according to claim 1, characterized in that: The establishment of the machine learning oxygen consumption prediction model includes the following steps: data processing, cleaning the collected vital sign data and oxygen pressure data; data modeling, using the vital sign data monitored by the smart watch as a feature variable for modeling; Data set division: the collected vital sign data and oxygen pressure data are divided into training set and test set. The training set is used to train the machine learning oxygen consumption prediction model, and the test set is used to evaluate the prediction performance of the trained model. Model parameter adjustment: various parameters in the machine learning oxygen consumption prediction model are adjusted to a reasonable range. Model correction: new vital sign data and oxygen pressure data are continuously introduced to optimize the model.

3. The method for adjusting an oxygen consumption regulating respirator based on machine learning of vital signs according to claim 2, characterized in that: In the data processing step of establishing the machine learning oxygen consumption prediction model, the missing values ​​and outliers in the collected vital sign data and oxygen pressure data are cleaned and deduplicated.

4. The method for adjusting an oxygen consumption regulating respirator based on machine learning of vital signs according to claim 3 is characterized in that: In the data modeling step of establishing a machine learning oxygen consumption prediction model, vital sign data include blood oxygen, heart rate, body temperature, and exercise status.

5. The method for adjusting an oxygen consumption regulating respirator based on machine learning of vital signs according to claim 4, characterized in that: The machine learning oxygen consumption prediction model includes multiple decision trees, each decision tree includes multiple sub-items, and different vital sign data are randomly extracted into each sub-item so that each decision tree can match a corresponding oxygen consumption.

6. An oxygen consumption regulating respirator based on machine learning of vital signs, characterized in that: The adjustment method used in any one of claims 1 to 5 comprises a smart watch, an oxygen cylinder, an airbag and an oxygen supply system connected between the oxygen cylinder and the airbag, wherein the oxygen supply system comprises an oxygen supply pipeline, a cooler, an inhalation hose, a connecting pipe, an exhalation hose and an absorption tank connected in sequence, wherein one end of the oxygen supply pipeline away from the cooler is connected to the oxygen cylinder, the cooler is also connected to the airbag, one end of the absorption tank away from the exhalation hose is connected to the airbag, and the connecting pipe is used to connect to a breathing mask.

7. The oxygen consumption regulating respirator based on machine learning of vital signs according to claim 6 is characterized in that: The oxygen supply pipeline is also provided with a pressure reducer and an intelligent regulating valve, and the intelligent regulating valve is located between the pressure reducer and the cooler.

8. The oxygen consumption regulating respirator based on machine learning of vital signs according to claim 7 is characterized in that: The intelligent regulating valve includes a valve body, a power supply fixed in the valve body, a motor, a microcontroller and a driving mechanism, the microcontroller is electrically connected to the motor, and the driving mechanism includes a shell fixed in the valve body, a base fixed under the shell, a positioning frame fixed in the shell, a rack slidably connected to the positioning frame, a gear meshing with the rack, a connecting column fixed at the lower end of the rack, a rubber pad fixed in the middle of the shell, a connecting rod fixed at the bottom of the rubber pad and a compression spring sleeved on the rod of the connecting rod, the bottom of the connecting column is fixed to the upper surface of the rubber pad; a blocking hole is vertically provided in the base, and connected air inlet and air outlet are respectively provided at both ends of the blocking hole, and the bottom end of the connecting rod can be inserted from the shell downward into the blocking hole to block the connection between the air inlet and the air outlet.

9. The oxygen consumption regulating respirator based on machine learning of vital signs according to claim 8, characterized in that: The oxygen pressure sensor is arranged on the air outlet, and the oxygen pressure sensor is electrically connected to the microcontroller.

10. The oxygen consumption regulating respirator based on machine learning of vital signs according to claim 9, characterized in that: The shell comprises an upper shell and a lower shell fixed together, wherein an annular protrusion is arranged at the inner connection of the upper shell and the lower shell, and the outer ring of the rubber pad is fixed on the annular protrusion and the gap is sealed with glue.

Images