Medicine cold chain aging optimization method and system based on neural network

Abstract

The invention relates to the technical field of medicine cold chains, in particular to a medicine cold chain aging optimization method and system based on a neural network, and the method comprises the steps: collecting and recording historical temperature data of medicine in a transportation process, collecting historical data of a medicine state, a transportation route and a traffic condition, and building a cold chain data set in combination with the historical temperature data; preprocessing the data in the cold chain data set, establishing a neural network model, and training the neural network model by using the preprocessed training set data; according to the method, related data of cold-chain logistics are collected, the related data comprise historical temperature data, medicine states, transportation routes, traffic conditions and the like, a neural network model is established by using historical data, the neural network model is trained, a real-time monitoring system is established, and the cold-chain logistics process is monitored in real time. And monitoring data is fed back to the neural network model, and a transportation plan and a transportation route are adjusted in time through real-time monitoring and feedback so as to deal with emergencies and optimize logistics timeliness.

Description

Technical Field

[0001] The present invention belongs to the field of pharmaceutical cold chain technology, and in particular relates to a pharmaceutical cold chain time optimization method and system based on a neural network. Background Art

[0002] Pharmaceutical cold chain items such as vaccines and biologics are sensitive to temperature and need to be stored and transported under specific temperature conditions. Once these drugs exceed the recommended storage temperature range, they may lose their activity or deteriorate, affecting efficacy and safety. Therefore, it is particularly important to control the timeliness of the pharmaceutical cold chain. The timeliness of the pharmaceutical cold chain refers to the time requirements for ensuring the quality and safety of drugs through strict temperature control measures in the production, storage, transportation, and use of drugs.

[0003] At present, the pharmaceutical cold chain is usually equipped with cold storage, refrigerated trucks, vehicle-mounted refrigerated boxes or insulated boxes and other cold chain equipment according to the scale and variety of operations, and ensure that the temperature automatic monitoring, display, recording, regulation and alarm functions of these equipment are normal.

[0004] However, since most companies mainly rely on manual detection and recording of temperature, the temperature control data is discontinuous, and there is a lack of technology for monitoring and feedback of the temperature of drugs throughout the process. When encountering emergencies, it is not convenient to adjust the transportation plan and route in time, which will extend the transportation time of the pharmaceutical cold chain and make the drugs at risk of deterioration or failure during transportation. Therefore, we need to propose a pharmaceutical cold chain time optimization method and system based on neural network to solve the above-mentioned problems, so that it can monitor and feedback in real time, adjust the transportation plan and route in time to respond to emergencies and optimize logistics timeliness, improve the efficiency of pharmaceutical cold chain transportation, and reduce the probability of drug deterioration or failure during transportation. Summary of the invention

[0005] In view of the above problems, the present invention provides a method for optimizing the timeliness of a pharmaceutical cold chain based on a neural network, comprising the following steps:

[0006] S1. Collect and record historical temperature data of medicines during transportation;

[0007] S2, collect historical data on medicine status, transportation routes and traffic conditions, and build a cold chain data set in combination with historical temperature data;

[0008] S3, preprocess the data in the cold chain dataset, and divide the processed data into training set, test set and validation set;

[0009] S4. Establish a neural network model based on the characteristics of cold chain logistics data and business needs;

[0010] S5. Use the preprocessed training set data to train the neural network model. When the performance of the neural network model reaches stability or optimality on the validation set, stop the training.

[0011] S6. Establish a real-time monitoring system on the cold chain equipment, and collect real-time data of the cold chain equipment through the real-time monitoring system;

[0012] S7, applying the real-time data to the trained neural network model, and predicting the emergency situation through the neural network model;

[0013] S71, when an emergency is predicted, the optimal transportation route is planned and the response strategy is formulated according to the prediction results and the actual situation, and then enter S8;

[0014] S72. When no emergency is predicted, transport according to the original transport route and return to S7;

[0015] S8. Evaluate the feasibility of the optimal route planning and response strategy. If feasible, make adjustments according to the feasible optimal route planning and response strategy. If not feasible, return to S71.

[0016] Furthermore, in step S2, by installing a GPS locator, a humidity sensor, a vibration sensor and a camera on the cold chain equipment, the GPS locator is used to track the position and driving trajectory of the transport vehicle in real time, the humidity sensor is used to monitor the humidity of the pharmaceutical environment to prevent the medicine from getting damp and deteriorating, and the vibration sensor is used to monitor whether the goods collide or tilt during transportation; the camera is used to monitor the status of the medicine in real time and record the specific conditions during transportation. The GPS locator, humidity sensor, vibration sensor and camera are installed to automatically record data every 5-10 seconds, and the recorded data is transmitted to the cloud storage in real time through the wireless communication module.

[0017] Furthermore, in step S3, the process of preprocessing the data in the cold chain data set is as follows:

[0018] S31, identifying outliers in the data of the cold chain data set to find out the outliers in the data;

[0019] S32. Process the abnormal data according to the abnormal value situation to make the data in the cold chain data set complete;

[0020] S33, converting the processed data into standard normal distribution data with a value of 0 and a standard deviation of 1;

[0021] S34. Extract characteristic data from the standard normal distribution data, and divide the data into a test set, a training set, and a validation set according to the characteristic data in proportions of 15%, 70%, and 15%.

[0022] Furthermore, in step S31, the process of outlier identification is as follows:

[0023] A1. Set the medicine storage temperature range. Data beyond the medicine storage temperature range is determined as abnormal value, and data within the medicine storage temperature range is determined as normal value;

[0024] A2. Set the humidity range for medicine storage. Data beyond the humidity range is considered abnormal, and data within the humidity range is considered normal.

[0025] A3. Set the geographic location fence range. When the cold chain equipment leaves the preset geographic location area, it is judged as an abnormal value. When the cold chain equipment is within the set geographic location, it is judged as a normal value.

[0026] A4. Set the thresholds for vibration intensity and vibration frequency. When the vibration intensity or vibration frequency exceeds the set threshold, it is judged as an abnormal value. When the vibration intensity and vibration frequency are both within the set threshold range, it is judged as a normal value.

[0027] A5. Set up video surveillance screen judgment rules. When the surveillance screen shows one or more of the following: obstruction, blur, black screen, illegal intrusion, and illegal movement of drugs, it is judged as an abnormal value, otherwise it is judged as a normal value.

[0028] Furthermore, in step S32, the process of processing abnormal data is as follows: temperature abnormal values ​​and humidity abnormal values ​​are directly deleted, and data that meets the set humidity range is inserted at the deleted position; geographic location abnormal values, vibration intensity and frequency abnormal values, and video surveillance screen abnormal values ​​are directly deleted.

[0029] Furthermore, in step S33, the process of converting the processed data into standard normal distribution data is as follows:

[0030] B1. Calculate the mean of each data type according to the data type in the cold chain dataset Mean The calculation formula is as follows:

[0031]

[0032]

[0033] Where n is the total number of samples, x i is the characteristic value of the i-th sample;

[0034] B2. According to the calculated mean Then calculate the standard deviation σ of each feature. The calculation formula of standard deviation σ is as follows:

[0035]

[0036] B3. Use the normal distribution conversion formula to convert the data into the standard normal distribution value Z according to each data point. The normal distribution conversion formula is as follows:

[0037]

[0038] B4. For video surveillance data, multiple features need to be merged using a merging formula and then proceed to step B5. For non-video data, proceed directly to step B5.

[0039] The combined formula is: Z = [Z1; Z2; ...; Z n ];

[0040] B5. Draw a histogram based on the converted data to check whether the converted data is a standard normal distribution. If it is a standard normal distribution, the conversion is completed. If it is not a standard normal distribution, the preprocessed data needs to be reconverted to a standard normal distribution value.

[0041] Furthermore, in step S4, the process of establishing the neural network model is as follows:

[0042] S41. Calculate the weight value w according to the amount of information carried by the cold chain data i , weight value w i The calculation formula is as follows:

[0043]

[0044] Among them, σ i is the standard deviation of the ith sample, σ j is the standard deviation of the jth sample, and n is the total number of samples;

[0045] S42, according to the calculated weight value w i , establish a neural network model, the neural network model establishment formula is as follows:

[0046]

[0047] Among them, x i is the input data, b i is the threshold set for the model, n is the total number of samples, a i is the bias term, Y i Output values ​​for the model;

[0048] S43, according to the model output value Y i Calculate the loss value L. The calculation formula of the loss value L is as follows:

[0049]

[0050] in, is the model prediction value, Y i is the actual value of the model, and n is the total number of samples;

[0051] S44. Update the weight value W using an iterative formula according to the loss value L to minimize the loss function. The iterative formula is as follows:

[0052]

[0053] Where η is the learning rate, is the model prediction value, Y i is the actual value of the model.

[0054] Furthermore, in step S5, when training the neural network model, the preprocessed training set data is input into the neural network model, and the weight values ​​w are calculated respectively. i , model output value Y i , loss value L and weight value W, and then calculate the proportion of the number of samples correctly predicted by the neural network model to the total number of samples based on the calculation results. When the proportion of the number of correctly predicted samples to the total number of samples exceeds 95%, the neural network model is judged to be stable or optimal, and training is stopped. Otherwise, the neural network model is judged to be unstable or poor, and training needs to continue.

[0055] Furthermore, in step S8, when evaluating the optimal route planning, the transportation time of each route is evaluated to ensure that the medicine can be delivered in the shortest time, which is feasible; otherwise, it is not feasible; when evaluating the response strategy, it is feasible if the medicine can be stably operated during transportation and the temperature and humidity are within the medicine storage range, otherwise, it is not feasible and needs to be adjusted according to the medicine storage environment.

[0056] Based on the above-described neural network-based pharmaceutical cold chain time-efficiency optimization method, the present invention also provides a neural network-based pharmaceutical cold chain time-efficiency optimization system, including: a data acquisition module, the data acquisition module collects and records historical temperature data of the pharmaceutical during transportation, as well as historical data of the pharmaceutical status, transportation route and traffic conditions;

[0057] A data preprocessing module, which preprocesses and divides the collected data;

[0058] A neural network model building and training module, which builds a neural network model according to the characteristics of cold chain logistics data and business needs, and uses the preprocessed data to train the neural network model;

[0059] A real-time data monitoring and acquisition module, which is used to establish a real-time monitoring system and collect real-time data of cold chain equipment through the real-time monitoring system;

[0060] A prediction module, which is used to apply real-time data to the trained neural network model to predict emergencies, and to plan optimal routes and formulate response strategies based on the predicted emergencies;

[0061] An execution module is used to evaluate the feasibility of the optimal route planning and response strategy, and make adjustments based on the feasible optimal route planning and response strategy.

[0062] The beneficial effects of the present invention are:

[0063] 1. The present invention collects relevant data of cold chain logistics, including historical temperature data, medical status, transportation routes, traffic conditions, etc., uses historical data to establish a neural network model, trains the neural network model, establishes a real-time monitoring system, monitors the cold chain logistics process in real time, and feeds the monitoring data back to the neural network model. Through real-time monitoring and feedback, the transportation plan and route are adjusted in time to deal with emergencies and optimize logistics timeliness.

[0064] 2. The present invention achieves the improvement of transportation efficiency, the guarantee of drug quality, the reduction of operating costs, the improvement of customer satisfaction and the enhancement of decision-making support capabilities through the cooperation of data acquisition module, data preprocessing module, real-time data monitoring and acquisition module, prediction module and execution module, thereby promoting the optimization of timeliness and reliability of pharmaceutical cold chain logistics.

[0065] Other features and advantages of the present invention will be described in the following description, and partly become apparent from the description, or understood by practicing the present invention. The purpose and other advantages of the present invention can be realized and obtained by the structures pointed out in the description, claims and drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0066] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0067] Figure 1 A flowchart according to an embodiment of the present invention is shown;

[0068] Figure 2 A flowchart of preprocessing data in a cold chain data set according to an embodiment of the present invention is shown;

[0069] Figure 3A schematic diagram of a process for outlier identification according to an embodiment of the present invention is shown;

[0070] Figure 4 A system block diagram according to an embodiment of the present invention is shown. DETAILED DESCRIPTION

[0071] In order to make the purpose, technical solution and advantages of the embodiments of the present invention clearer, the technical solution in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0072] The embodiment of the present invention provides a method for optimizing the timeliness of a pharmaceutical cold chain based on a neural network, such as Figure 1-4 As shown, the following steps are included:

[0073] S1. Collect and record historical temperature data of medicines during transportation;

[0074] The temperature sensors installed on the cold chain equipment collect and record the temperature data of the medicines in the transportation process in real time. The temperature data includes the temperature change trend and fluctuation range.

[0075] S2, collect historical data on medicine status, transportation routes and traffic conditions, and build a cold chain data set in combination with historical temperature data;

[0076] By installing GPS locators, humidity sensors, vibration sensors and cameras on cold chain equipment, GPS locators are used to track the location and driving trajectory of transport vehicles in real time, humidity sensors are used to monitor the humidity of the pharmaceutical environment to prevent the pharmaceuticals from getting damp and deteriorating, and vibration sensors are used to monitor whether the goods collide or tilt during transportation; cameras are used to monitor the status of pharmaceuticals in real time and record specific conditions during transportation. GPS locators, humidity sensors, vibration sensors and cameras are installed to automatically record data every 5-10 seconds, and the recorded data is transmitted to cloud storage in real time through wireless communication modules;

[0077] When the cold chain database is established, the temperature data, drug status, traffic conditions and transportation route data are classified and integrated, and a cold chain data set is established based on the classified and integrated data; the drug status includes whether it is damaged or deteriorated; the transportation route includes the starting point, the end point and the passing place information; the traffic conditions include road congestion and weather conditions;

[0078] S3, preprocess the data in the cold chain dataset, and divide the processed data into training set, test set and validation set;

[0079] like Figure 2 As shown in the figure, the process of preprocessing the data in the cold chain dataset is as follows:

[0080] S31, identifying outliers in the data of the cold chain data set to find out the outliers in the data;

[0081] like Figure 3 As shown in Figure 2, the process of outlier identification is as follows:

[0082] A1. Set the medicine storage temperature range. Data beyond the medicine storage temperature range is determined as abnormal value, and data within the medicine storage temperature range is determined as normal value;

[0083] A2. Set the humidity range for medicine storage. Data beyond the humidity range is considered abnormal, and data within the humidity range is considered normal.

[0084] A3. Set the geographic location fence range. When the cold chain equipment leaves the preset geographic location area, it is judged as an abnormal value. When the cold chain equipment is within the set geographic location, it is judged as a normal value.

[0085] A4. Set the thresholds for vibration intensity and vibration frequency. When the vibration intensity or vibration frequency exceeds the set threshold, it is judged as an abnormal value. When the vibration intensity and vibration frequency are both within the set threshold range, it is judged as a normal value.

[0086] A5. Set the video surveillance screen judgment rules. When the surveillance screen shows one or more of the following: occlusion, blur, black screen, illegal intrusion, and illegal movement of drugs, it is judged as an abnormal value, otherwise it is judged as a normal value;

[0087] S32. Process the abnormal data according to the abnormal value situation to make the data in the cold chain data set complete;

[0088] The process of handling abnormal data is as follows: directly delete abnormal temperature and humidity values, and insert data that meets the set humidity range in the deleted position; directly delete abnormal geographical location values, abnormal vibration intensity and frequency values, and abnormal video surveillance images.

[0089] S33, converting the processed data into standard normal distribution data with a value of 0 and a standard deviation of 1;

[0090] The process of converting the processed data into standard normal distribution data is as follows:

[0091] B1. Calculate the mean of each data type according to the data type in the cold chain dataset Mean The calculation formula is as follows:

[0092]

[0093] Where n is the total number of samples, x i is the characteristic value of the i-th sample;

[0094] B2. According to the calculated mean Then calculate the standard deviation σ of each feature. The calculation formula of standard deviation σ is as follows:

[0095]

[0096] B3. Use the normal distribution conversion formula to convert the data into the standard normal distribution value Z according to each data point. The normal distribution conversion formula is as follows:

[0097]

[0098] B4. For video surveillance data, multiple features need to be merged using a merging formula and then proceed to step B5. For non-video data, proceed directly to step B5.

[0099] The combined formula is: Z = [Z1; Z2; ...; Z n ];

[0100] B5. Draw a histogram based on the converted data to check whether the converted data is a standard normal distribution. If it is a standard normal distribution, the conversion is completed. If it is not a standard normal distribution, the preprocessed data needs to be reconverted to a standard normal distribution value.

[0101] S34, extracting characteristic data from the standard normal distribution data, and dividing the test set, the training set, and the validation set into a ratio of 15%, 70%, and 15% according to the characteristic data;

[0102] S4. Establish a neural network model based on the characteristics of cold chain logistics data and business needs;

[0103] The process of establishing a neural network model is as follows:

[0104] S41. Calculate the weight value w according to the amount of information carried by the cold chain data i , weight value w i The calculation formula is as follows:

[0105]

[0106] Among them, σ i is the standard deviation of the ith sample, σ j is the standard deviation of the jth sample, and n is the total number of samples;

[0107] S42, according to the calculated weight value w i , establish a neural network model, the neural network model establishment formula is as follows:

[0108]

[0109] Among them, x i is the input data, b i is the threshold set for the model, n is the total number of samples, a i is the bias term, Y i Output values ​​for the model;

[0110] S43, according to the model output value Y i Calculate the loss value L. The calculation formula of the loss value L is as follows:

[0111]

[0112] in, is the model prediction value, Y i is the actual value of the model, and n is the total number of samples;

[0113] S44. Update the weight value W using an iterative formula according to the loss value L to minimize the loss function. The iterative formula is as follows:

[0114]

[0115] Where η is the learning rate, is the model prediction value, Y i is the actual value of the model;

[0116] S5. Use the preprocessed training set data to train the neural network model. When the performance of the neural network model reaches stability or optimality on the validation set, stop the training.

[0117] When training the neural network model, the preprocessed training set data is input into the neural network model, and the weight values ​​w are calculated respectively. i , model output value Y i , loss value L and weight value W, and then calculate the proportion of samples correctly predicted by the neural network model to the total number of samples based on the calculation results. When the proportion of correctly predicted samples to the total number of samples exceeds 95%, the neural network model is determined to be stable or optimal, and training is stopped. Otherwise, the neural network model is determined to be unstable or poor, and training needs to be continued;

[0118] S6. Establish a real-time monitoring system on the cold chain equipment, and collect real-time data of the cold chain equipment through the real-time monitoring system;

[0119] The cold chain equipment includes a transport vehicle and a medicine storage box installed on the transport vehicle. The real-time monitoring system includes a camera, a temperature and humidity sensor, a GPS locator, a vibration sensor and a smoke sensor. Cameras are installed inside the medicine storage box and on the front of the transport vehicle. The GPS locator and the smoke sensor are installed on the transport vehicle. The vibration sensor and the temperature and humidity sensor are installed inside the medicine storage box. The medicine storage box is also equipped with a temperature control mechanism for controlling the temperature and humidity inside the medicine storage box.

[0120] S7, applying the real-time data to the trained neural network model, and predicting the emergency situation through the neural network model;

[0121] S71, when an emergency is predicted, the optimal transportation route is planned and the response strategy is formulated according to the prediction results and the actual situation, and then enter S8;

[0122] S72. When no emergency is predicted, transport according to the original transport route and return to S7;

[0123] S8. Evaluate the feasibility of the optimal route planning and response strategy. If feasible, make adjustments according to the feasible optimal route planning and response strategy. If not feasible, return to S71.

[0124] When evaluating the optimal route planning, the transportation time of each route is evaluated to ensure that the medicine can be delivered in the shortest time. If so, it is feasible; otherwise, it is not feasible. When evaluating the response strategy, it is feasible if the medicine can be able to operate stably during transportation and the temperature and humidity are within the storage range of the medicine. Otherwise, it is not feasible and needs to be adjusted according to the medicine storage environment.

[0125] By collecting relevant data on cold chain logistics, including historical temperature data, medical status, transportation routes, traffic conditions, etc., and using historical data to establish a neural network model, and training the neural network model, a real-time monitoring system is established to monitor the cold chain logistics process in real time, and the monitoring data is fed back to the neural network model. Through real-time monitoring and feedback, transportation plans and routes can be adjusted in time to respond to emergencies and optimize logistics timeliness.

[0126] Based on the above-described neural network-based pharmaceutical cold chain time-efficiency optimization method, the present invention also provides a neural network-based pharmaceutical cold chain time-efficiency optimization system, such as Figure 4 As shown, it includes: a data acquisition module, which collects and records historical temperature data of medicines during transportation, as well as historical data of medicine status, transportation route and traffic conditions;

[0127] A data preprocessing module, which preprocesses and divides the collected data;

[0128] A neural network model building and training module, which builds a neural network model according to the characteristics of cold chain logistics data and business needs, and uses the preprocessed data to train the neural network model;

[0129] A real-time data monitoring and acquisition module, which is used to establish a real-time monitoring system and collect real-time data of cold chain equipment through the real-time monitoring system;

[0130] A prediction module, which is used to apply real-time data to the trained neural network model to predict emergencies, and to plan optimal routes and formulate response strategies based on the predicted emergencies;

[0131] An execution module is used to evaluate the feasibility of the optimal route planning and response strategy, and make adjustments based on the feasible optimal route planning and response strategy.

[0132] Through the cooperation of data acquisition module, data preprocessing module, real-time data monitoring and acquisition module, prediction module and execution module, the transportation efficiency is improved, the quality of drugs is guaranteed, the operating cost is reduced, the customer satisfaction is improved, and the decision-making support capability is enhanced, which promotes the optimization of the timeliness and reliability of pharmaceutical cold chain logistics.

[0133] Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent substitutions for some of the technical features therein; and these modifications or substitutions do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for optimizing the timeliness of a pharmaceutical cold chain based on a neural network, characterized in that: The following steps are involved: S1. Collect and record historical temperature data of medicines during transportation; S2, collect historical data on medicine status, transportation routes and traffic conditions, and build a cold chain data set in combination with historical temperature data; S3, preprocess the data in the cold chain dataset, and divide the processed data into training set, test set and validation set; S4. Establish a neural network model based on the characteristics of cold chain logistics data and business needs; S5. Use the preprocessed training set data to train the neural network model. When the performance of the neural network model reaches stability or optimality on the validation set, stop the training. S6. Establish a real-time monitoring system on the cold chain equipment, and collect real-time data of the cold chain equipment through the real-time monitoring system; S7, applying the real-time data to the trained neural network model, and predicting emergencies through the neural network model; S71, when an emergency is predicted, the optimal transportation route is planned and the response strategy is formulated according to the prediction results and the actual situation, and then enter S8; S72. When no emergency is predicted, transport according to the original transport route and return to S7; S8. Evaluate the feasibility of the optimal route planning and response strategy. If feasible, make adjustments according to the feasible optimal route planning and response strategy. If not feasible, return to S71.

2. The method for optimizing the pharmaceutical cold chain time efficiency based on a neural network according to claim 1, characterized in that: In step S2, by installing a GPS locator, a humidity sensor, a vibration sensor and a camera on the cold chain equipment, the GPS locator is used to track the position and driving trajectory of the transport vehicle in real time, the humidity sensor is used to monitor the humidity of the pharmaceutical environment to prevent the medicine from getting damp and deteriorating, and the vibration sensor is used to monitor whether the goods collide or tilt during transportation; the camera is used to monitor the status of the medicine in real time and record the specific conditions during transportation. The GPS locator, humidity sensor, vibration sensor and camera are installed to automatically record data every 5-10 seconds, and the recorded data is transmitted to the cloud storage in real time through the wireless communication module.

3. The method for optimizing the pharmaceutical cold chain time efficiency based on a neural network according to claim 2, characterized in that: In step S3, the process of preprocessing the data in the cold chain data set is as follows: S31, identifying outliers in the data of the cold chain data set to find out the outliers in the data; S32. Process the abnormal data according to the abnormal value situation to make the data in the cold chain data set complete; S33, converting the processed data into standard normal distribution data with a value of 0 and a standard deviation of 1; S34. Extract characteristic data from the standard normal distribution data, and divide the data into a test set, a training set, and a validation set according to the characteristic data in proportions of 15%, 70%, and 15%.

4. The method for optimizing the pharmaceutical cold chain time efficiency based on a neural network according to claim 3, characterized in that: In step S31, the process of outlier identification is as follows: A1. Set the medicine storage temperature range. Data beyond the medicine storage temperature range is determined as abnormal value, and data within the medicine storage temperature range is determined as normal value; A2. Set the humidity range for medicine storage. Data beyond the humidity range is considered abnormal, and data within the humidity range is considered normal. A3. Set the geographic location fence range. When the cold chain equipment leaves the preset geographic location area, it is judged as an abnormal value. When the cold chain equipment is within the set geographic location, it is judged as a normal value. A4. Set the thresholds for vibration intensity and vibration frequency. When the vibration intensity or vibration frequency exceeds the set threshold, it is judged as an abnormal value. When the vibration intensity and vibration frequency are both within the set threshold range, it is judged as a normal value. A5. Set up video surveillance screen judgment rules. When the surveillance screen shows one or more of the following: obstruction, blur, black screen, illegal intrusion, and illegal movement of drugs, it is judged as an abnormal value, otherwise it is judged as a normal value.

5. The method for optimizing the pharmaceutical cold chain time efficiency based on a neural network according to claim 4, characterized in that: In step S32, the process of processing abnormal data is as follows: the temperature abnormal values ​​and humidity abnormal values ​​are directly deleted, and data that meets the set humidity range is inserted at the deleted position; the geographical location abnormal values, vibration intensity and frequency abnormal values, and video surveillance screen abnormal values ​​are directly deleted.

6. The method for optimizing the pharmaceutical cold chain time efficiency based on a neural network according to claim 5, characterized in that: In step S33, the process of converting the processed data into standard normal distribution data is as follows: B1. Calculate the mean of each data type according to the data type in the cold chain dataset Mean The calculation formula is as follows: Where n is the total number of samples, x i is the characteristic value of the i-th sample; B2. According to the calculated mean Then calculate the standard deviation σ of each feature. The calculation formula of standard deviation σ is as follows: B3. Use the normal distribution conversion formula to convert the data into the standard normal distribution value Z according to each data point. The normal distribution conversion formula is as follows: B4. For video surveillance data, multiple features need to be merged using a merging formula and then proceed to step B5. For non-video data, proceed directly to step B5. The combined formula is: Z = [Z1; Z2; ...; Z n ]; B5. Draw a histogram based on the converted data to check whether the converted data is a standard normal distribution. If it is a standard normal distribution, the conversion is completed. If it is not a standard normal distribution, the preprocessed data needs to be reconverted to a standard normal distribution value.

7. The method for optimizing the pharmaceutical cold chain time efficiency based on a neural network according to claim 6, characterized in that: In step S4, the process of establishing the neural network model is as follows: S41. Calculate the weight value w according to the amount of information carried by the cold chain data i , weight value w i The calculation formula is as follows: Among them, σ i is the standard deviation of the ith sample, σ j is the standard deviation of the jth sample, and n is the total number of samples; S42, according to the calculated weight value w i , establish a neural network model, the neural network model establishment formula is as follows: Among them, x i is the input data, b i is the threshold set for the model, n is the total number of samples, a i is the bias term, Y i Output values ​​for the model; S43, according to the model output value Y i Calculate the loss value L. The calculation formula of the loss value L is as follows: in, is the model prediction value, Y i is the actual value of the model, n is the total number of samples; S44. Update the weight value W using an iterative formula according to the loss value L to minimize the loss function. The iterative formula is as follows: Where η is the learning rate, is the model prediction value, Y i is the actual value of the model.

8. The method for optimizing the pharmaceutical cold chain time efficiency based on a neural network according to claim 7, characterized in that: In step S5, when training the neural network model, the preprocessed training set data is input into the neural network model, and the weight values ​​w are calculated respectively. i , model output value Y i , loss value L and weight value W, and then calculate the proportion of the number of samples correctly predicted by the neural network model to the total number of samples based on the calculation results. When the proportion of the number of correctly predicted samples to the total number of samples exceeds 95%, the neural network model is judged to be stable or optimal, and training is stopped. Otherwise, the neural network model is judged to be unstable or poor, and training needs to continue.

9. The method for optimizing the pharmaceutical cold chain time efficiency based on a neural network according to claim 8, characterized in that: In step S8, when evaluating the optimal route planning, the transportation time of each route is evaluated to ensure that the medicine can be delivered in the shortest time, which is feasible; otherwise, it is not feasible; when evaluating the response strategy, it is feasible if the medicine can be stably operated during transportation and the temperature and humidity are within the medicine storage range, otherwise, it is not feasible and needs to be adjusted according to the medicine storage environment.

10. A medicine cold chain time optimization system based on a neural network, based on a medicine cold chain time optimization method based on a neural network according to any one of claims 1 to 9, characterized in that: include: A data collection module, which collects and records historical temperature data of medicines during transportation, as well as historical data of medicine status, transportation routes and traffic conditions; A data preprocessing module, which preprocesses and divides the collected data; A neural network model building and training module, which builds a neural network model according to the characteristics of cold chain logistics data and business needs, and uses the preprocessed data to train the neural network model; A real-time data monitoring and acquisition module, which is used to establish a real-time monitoring system and collect real-time data of cold chain equipment through the real-time monitoring system; A prediction module, which is used to apply real-time data to the trained neural network model to predict emergencies, and to plan optimal routes and formulate response strategies based on the predicted emergencies; An execution module is used to evaluate the feasibility of the optimal route planning and response strategy, and make adjustments based on the feasible optimal route planning and response strategy.

Images