A container environment spatiotemporal status early warning method

Abstract

The present invention belongs to the field of environmental state early warning technology, and specifically relates to a container environment spatiotemporal state early warning method, comprising: S1: constructing a data sample set; S2: constructing a training sample set; S3: training a regression model: using the multiple training sample sets obtained in S2 to perform SVR model training to obtain multiple SVR regression models; S4: determining whether the container spatiotemporal state is abnormal. The method of the present invention eliminates the need for a large number of sensors, and instead uses multiple SVR regression models to verify each other to determine whether the spatiotemporal state of the container environment is abnormal. This method can improve the accuracy of container spatiotemporal state abnormality determinations, resolving the problem that existing containers only have sensor sensors installed in one location, making it impossible to monitor spatiotemporal state abnormalities within the container environment.

Description

Technical Field

[0001] The present invention belongs to the technical field of environmental status early warning, and in particular relates to a container environmental spatiotemporal status early warning method. Background Art

[0002] A container is a large cargo container with specific strength, rigidity, and specifications designed for circulation. With the booming global trade, containers have become a crucial means of transporting goods. Currently, containerized transport is becoming a growing trend. As the primary enclosed freight carrier, the internal environment of a container significantly impacts the goods stored. During long-distance land and water transport, changes in the external environment can significantly impact the container's internal environment. In actual transportation, the container's internal environment can vary significantly between daytime and nighttime conditions. Furthermore, the cargo's ability to absorb and release ambient heat can also significantly affect the internal environment. In actual production, the container's internal environment can significantly impact the goods, potentially even affecting their quality. For certain flammable goods, the elevated container temperature can cause fires. Therefore, when using containers to transport goods, the environmental conditions such as temperature, humidity, and pressure inside the container are generally monitored. However, due to the large size of the container itself, the sensing sensor is generally only installed in one place during monitoring, which makes it impossible to directly obtain the overall environmental status of the container. When the spatiotemporal state of the environment inside the container is abnormal, it cannot be monitored. Therefore, it is urgent to monitor the overall environment inside the container, promptly detect abnormal spatiotemporal state of the environment inside the container and issue an early warning, so as to improve the safety of container transportation. Summary of the Invention

[0003] The present invention solves the problem that existing containers only have a sensing sensor installed in one place and cannot monitor abnormal spatiotemporal conditions of the environment inside the container. It provides a container environment spatiotemporal condition early warning method, which monitors the overall environmental condition inside the container and issues an early warning when the environmental condition inside the container becomes abnormal.

[0004] The technical solutions claimed in the present invention are as follows:

[0005] A container environment spatiotemporal state early warning method comprises the following steps:

[0006] S1: Constructing a data sample set: Constructing a spatiotemporal data sample set for identifying abnormalities in the container logistics transportation environment, wherein the data sample set includes the collected spatiotemporal environmental status data and the time when the status was collected; the spatiotemporal environmental status data includes multiple types;

[0007] S2: Constructing training sample sets: Constructing multiple training sample sets, each training sample set corresponding to a certain type of spatiotemporal environmental state data; including: traversing the spatiotemporal data sample set in S1, then traversing the spatiotemporal environmental state data of the samples in the spatiotemporal data sample set, using the value corresponding to the currently traversed spatiotemporal environmental state data as output data, and the values ​​corresponding to other spatiotemporal environmental states as input data, combining the input data and the output data into a set of data, and storing them in the corresponding training sample set; the number of the training sample sets is equal to the number of types of spatiotemporal environmental state data in S1;

[0008] S3: training regression model: using the multiple training sample sets obtained in S2 to perform SVR model training to obtain multiple SVR regression models; finding the minimum model error corresponding to each of the SVR regression models; each of the SVR regression models is trained on the spatiotemporal environmental state data corresponding to the training sample set;

[0009] S4: Determine whether the spatiotemporal state of the container is abnormal: collect the current spatiotemporal environmental state data of the container in real time, and input the current spatiotemporal environmental state data of the container into the corresponding SVR regression model in S3 respectively. If the output of one of the SVR regression models exceeds the minimum model error corresponding to the model compared with the spatiotemporal environmental state, the cumulative number of models that exceed the minimum model error is calculated. If the number of models that exceed the minimum model error reaches a predetermined number of times, it is considered that the spatiotemporal environmental state corresponding to the SVR regression model is abnormal.

[0010] Preferably, the spatiotemporal environmental status data in S1 is collected by a perception sensor installed in the container.

[0011] Preferably, the spatial and temporal environmental status data in the container includes temperature, pressure and humidity.

[0012] Preferably, the method for calculating the minimum model error in S3 is to input the input data of each group of data in the training sample set of the SVR regression model obtained by training S3 into the corresponding SVR regression model in sequence, obtain the output result, calculate the absolute value of the difference between the output result and the output data in this group of data, and the smallest value among all absolute values ​​is the minimum model error of the corresponding SVR regression model.

[0013] Preferably, in S4, perception sensors installed in the container are used to collect the current spatiotemporal environment status data in the container in real time.

[0014] Preferably, the predetermined number of times in S4 is set according to actual needs.

[0015] Preferably, the container is provided with a container environment spatiotemporal state early warning system, comprising: a temperature control unit, a humidity control unit, and a pressure control unit, and the main control unit is interconnected with the temperature control unit, the humidity control unit, and the pressure control unit;

[0016] The temperature control unit includes a refrigerator and a temperature sensor, which senses the temperature inside the container and transmits the sensed temperature information to the main control unit;

[0017] The humidity control unit includes a humidity adjustment device and a humidity sensor, which senses the humidity inside the container and transmits the sensed humidity information to the main control unit;

[0018] The pressure regulating unit includes a pressure sensor and two sets of pressure regulating devices, which respectively increase and reduce pressure. The pressure sensor senses the pressure inside the container and transmits the sensed pressure value to the main control unit.

[0019] Preferably, in S4, if a certain model is always abnormal under the continuous state acquisition condition, it is considered that the state needs to be adjusted, and the main control unit controls the corresponding temperature control unit, humidity control unit or pressure control unit to intervene.

[0020] Beneficial effects:

[0021] The present invention provides a container environment spatiotemporal state early warning method, which constructs a spatiotemporal data sample set for identifying abnormalities in a container logistics and transportation environment, wherein the data sample set includes collected spatiotemporal environment state data and the time when the state is collected, and the spatiotemporal environment state data includes multiple types; constructs a training sample set, and according to the type of the spatiotemporal environment state data, the spatiotemporal data sample set is divided into multiple training sample sets, each training sample set includes multiple groups of input data and output data, wherein the output data is a certain spatiotemporal environment state data of the container, and the input is other spatiotemporal environment state data of the container; then, the multiple training sample sets are used to respectively perform SVR model training to obtain multiple SVR regression models, and the abnormal situation of the container is judged only based on the container state sensed by a sensor installed at one place; the method of the present invention does not require a large number of sensors to be arranged, but uses multiple SVR regression models to verify each other to judge the abnormality of the spatiotemporal state of the environment in the container, and can improve the accuracy of the judgment of the abnormality of the spatiotemporal state of the environment in the container, thereby solving the problem that the existing container only has a perception sensor installed at one place and cannot monitor the abnormality of the spatiotemporal state of the environment in the container. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] Figure 1 It is a flow chart of the present invention. DETAILED DESCRIPTION

[0023] The present invention will be further described below with reference to the accompanying drawings:

[0024] A container environment spatiotemporal state early warning method, such as Figure 1 As shown, it includes the following steps:

[0025] S1: Constructing a data sample set: constructing a spatiotemporal data sample set for identifying abnormalities in container logistics and transportation environment, the data sample set includes the collected spatiotemporal environmental status data and the collection time of the status; the spatiotemporal environmental status data includes multiple types; the spatiotemporal environmental status data is collected by sensing sensors installed in the container; in a specific embodiment of the present invention, constructing a spatiotemporal data sample set X(t) for identifying abnormalities in container logistics and transportation environment, the data set includes the collected spatiotemporal status data and the collection time of the status, and the collected n types of environmental status include but are not limited to the temperature T, pressure P, and humidity H of the container.

[0026] S2: Constructing a training sample set: Constructing multiple training sample sets, each training sample set corresponding to a certain type of spatiotemporal environment state data; including: traversing the spatiotemporal data sample set in S1, then traversing the spatiotemporal environment state data of the samples in the spatiotemporal data sample set, taking the value corresponding to the currently traversed spatiotemporal environment state data as output data, and the values ​​corresponding to other spatiotemporal environment states as input data, forming a set of data with the input data and the output data, and storing it in the corresponding training sample set; the number of the training sample sets is equal to the number of types of the spatiotemporal environment state data in S1; in a specific embodiment of the present invention, including: (1) Constructing a training sample set X i , the sample is taken from the i-th sample in the spatiotemporal data sample set X(t), i is a natural number starting from 1, and i is not greater than n, and the spatiotemporal environment state corresponding to i is taken as the output y i , other spatiotemporal environment states as input x i , forming a training sample set X0=(y i (t),x i (t)); (2) if i is not greater than n, then take i=i+1 and repeat step (1); otherwise, stop; n is the number of samples in the spatiotemporal data sample set.

[0027] S3: training regression model: using the multiple training sample sets obtained in S2 to perform SVR model training to obtain multiple SVR regression models; finding the minimum model error corresponding to each of the SVR regression models; each of the SVR regression models is trained on the spatiotemporal environmental state data corresponding to the training sample set;

[0028] The method for calculating the minimum model error is to input the input data of each group of data in the training sample set used to train the SVR regression model obtained in S3 into the corresponding SVR regression model in sequence, obtain the output result, calculate the absolute value of the difference between the output result and the output data in this group of data, and the minimum value among all absolute values ​​is the minimum model error of the corresponding SVR regression model. In a specific embodiment of the present invention, the SVR model is used to train the regression model SVR i , thus obtaining the SVR model with the i-th state as the output, that is, z i =f(x i (t)), and the minimum model error α can be obtained i So that it satisfies max|y i (t)-z i (t)|≤α i ; where i starts from 1 and goes to n, where n is the number of samples in the corresponding training sample set.

[0029] S4: Determine whether the spatiotemporal state of the container is abnormal: Use the perception sensor installed in the container to collect the current spatiotemporal state data of the container in real time, and input the current spatiotemporal state data of the container into the corresponding SVR regression model in S3 respectively. If the output of a certain SVR regression model exceeds the minimum model error corresponding to the spatiotemporal state compared with the spatiotemporal state, and the cumulative number of models that exceed the minimum model error, if the number of models that exceed the minimum model error reaches a predetermined number of times, then it is considered that the spatiotemporal state corresponding to the SVR regression model is abnormal; the predetermined number of times can be set according to actual needs. In a specific embodiment of the present invention, the current spatiotemporal state of the container environment is collected in real time, and the n spatiotemporal state warning models are respectively brought into the model input to obtain the model output z i , if the model output is greater than the actual state than the model error α corresponding to the model i At this time, the cumulative number of models that exceed the model error is m. If the number of times the model error is exceeded reaches the predetermined number m, the spatiotemporal state of the container is considered to be abnormal.

[0030] In a specific embodiment of the present invention, the container is provided with a container environment spatiotemporal state early warning system, comprising: a temperature control unit, a humidity control unit, and a pressure control unit, wherein the main control unit is interconnected with the temperature control unit, the humidity control unit, and the pressure control unit;

[0031] The temperature control unit includes a refrigerator and a temperature sensor, which senses the temperature inside the container and transmits the sensed temperature information to the main control unit;

[0032] The humidity control unit includes a humidity adjustment device and a humidity sensor, which senses the humidity inside the container and transmits the sensed humidity information to the main control unit;

[0033] The pressure regulating unit includes a pressure sensor and two sets of pressure regulating devices, which respectively increase and reduce pressure. The pressure sensor senses the pressure inside the container and transmits the sensed pressure value to the main control unit.

[0034] If a model remains abnormal during continuous state acquisition in S4, it is considered that the state requires adjustment, and the main control unit controls the corresponding temperature control unit, humidity control unit, or pressure control unit to intervene. In a specific embodiment of the present invention, if the temperature inside the container is abnormal, such as too high or too low, the main control unit can control the activation of the temperature control unit to adjust the temperature inside the container.

Claims

1. A container environment spatiotemporal status early warning method, characterized in that: The steps include: S1: Constructing a data sample set: Constructing a spatiotemporal data sample set for identifying abnormalities in the container logistics transportation environment, wherein the data sample set includes the collected spatiotemporal environmental status data and the time when the status was collected; the spatiotemporal environmental status data includes multiple types; S2: Constructing training sample sets: Constructing multiple training sample sets, each training sample set corresponding to a certain type of spatiotemporal environmental state data; including: traversing the spatiotemporal data sample set in S1, then traversing the spatiotemporal environmental state data of the samples in the spatiotemporal data sample set, using the value corresponding to the currently traversed spatiotemporal environmental state data as output data, and the values ​​corresponding to other spatiotemporal environmental states as input data, combining the input data and the output data into a set of data, and storing them in the corresponding training sample set; the number of the training sample sets is equal to the number of types of spatiotemporal environmental state data in S1; S3: training regression model: using the multiple training sample sets obtained in S2 to perform SVR model training to obtain multiple SVR regression models; finding the minimum model error corresponding to each of the SVR regression models; each of the SVR regression models is trained on the spatiotemporal environmental state data corresponding to the training sample set; S4: Determine whether the spatiotemporal state of the container is abnormal: collect the current spatiotemporal environmental state data of the container in real time, and input the current spatiotemporal environmental state data of the container into the corresponding SVR regression model in S3 respectively. If the output of one of the SVR regression models exceeds the minimum model error corresponding to the model compared with the spatiotemporal environmental state, the cumulative number of models that exceed the minimum model error is calculated. If the number of models that exceed the minimum model error reaches a predetermined number of times, it is considered that the spatiotemporal environmental state corresponding to the SVR regression model is abnormal.

2. The container environment spatiotemporal state early warning method according to claim 1, characterized in that: The spatiotemporal environmental status data described in S1 is collected by the perception sensors installed in the container.

3. The container environment spatiotemporal state early warning method according to claim 2, characterized in that: The spatial and temporal environmental status data in the container include temperature, pressure and humidity.

4. The container environment spatiotemporal state early warning method according to claim 1, characterized in that: The method for calculating the minimum model error in S3 is to input the input data of each group of data in the training sample set of the SVR regression model obtained by S3 into the corresponding SVR regression model in turn, obtain the output result, calculate the absolute value of the difference between the output result and the output data in this group of data, and the smallest value among all absolute values ​​is the minimum model error of the corresponding SVR regression model.

5. The container environment spatiotemporal state early warning method according to claim 1, characterized in that: In S4, perception sensors installed in the container are used to collect the current spatiotemporal environmental status data in the container in real time.

6. The container environment spatiotemporal state early warning method according to claim 1, characterized in that: The predetermined number of times in S4 is set according to actual needs.

7. The container environment spatiotemporal state early warning method according to claim 2, characterized in that: The container is provided with a container environment spatiotemporal state early warning system, comprising: a temperature control unit, a humidity control unit, and a pressure control unit, wherein a main control unit is interconnected with the temperature control unit, the humidity control unit, and the pressure control unit; The temperature control unit includes a refrigerator and a temperature sensor, which senses the temperature inside the container and transmits the sensed temperature information to the main control unit; The humidity control unit includes a humidity adjustment device and a humidity sensor, which senses the humidity inside the container and transmits the sensed humidity information to the main control unit; The pressure regulating unit includes a pressure sensor and two sets of pressure regulating devices, which respectively increase and reduce pressure. The pressure sensor senses the pressure inside the container and transmits the sensed pressure value to the main control unit.

8. The container environment spatiotemporal state early warning method according to claim 7, characterized in that: In S4, if a model is always abnormal under the continuous state acquisition condition, it is considered that the state needs to be adjusted, and the main control unit controls the corresponding temperature control unit, humidity control unit or pressure control unit to intervene.

Images