A fast estimation method for container environmental parameters

Abstract

The invention belongs to the field of computing, and specifically relates to a method for quickly estimating container environmental parameters, comprising the following steps: S1: performing finite element simulation on the environment inside a container to obtain simulation data; S2: training a neural network model of the environment inside the container; S3: estimating the range occupied by items in the actual container; S4: matching the neural network model; S5: obtaining state parameters of items inside the container; S6: predicting the environmental parameters inside the container; using finite element simulation to obtain simulation data to establish a neural network model of the environmental parameters inside the container, using online data of the location of the sensor inside the container to estimate the environmental parameters inside the container, and on this basis using the neural network model to quickly obtain the environmental parameters of each location in the container, solving the problem in the prior art that the overall environmental state inside the container cannot be directly obtained due to the installation of a perception sensor at only one location in the container, and facilitating the monitoring of the environmental parameters inside the container.

Description

Technical Field

[0001] The invention belongs to the field of computing, and in particular relates to a method for quickly estimating container environmental parameters. Background Art

[0002] As the main carrier of closed freight, the internal environment of the container has an important impact on the goods stored. During long-distance land and water transportation, the external environment varies greatly, which will have a great impact on the environment inside the container. In fact, the environment inside the container will also change greatly during the day and night with sufficient sunshine. At the same time, the ability of the goods in the container to absorb and release environmental heat will also play an important role in the environment inside the container. Therefore, when using containers to transport important goods, the temperature, humidity, pressure and other environmental conditions in the container are generally monitored. However, due to the large size of the container itself, the sensing sensor is generally installed in only one place during monitoring, which makes it impossible to directly obtain the overall environmental status of the container; the environmental status includes temperature, pressure and humidity. Summary of the invention

[0003] The present invention solves the problem in the prior art that the overall environmental status in the container cannot be directly obtained because the perception sensor is only installed in one place in the container, and provides a method for quickly estimating container environmental parameters. The environmental status at any position in the entire container space can be estimated under the condition of installing only one sensor, thereby providing a reference for container environmental monitoring.

[0004] The technical solution claimed in the present invention is as follows:

[0005] A method for quickly estimating container environmental parameters comprises the following steps:

[0006] S1: Perform finite element simulation on the environment inside the container to obtain simulation data: divide the internal space of the container into three-dimensional grids and establish a grid coordinate system; set the external environment parameters of the container; set the spatial position and size of the items inside the container; set the state parameters of the items inside the container; perform finite element simulation on the temperature, pressure and humidity of the internal environment of the container to obtain the values ​​of the environmental parameters in all grids; store the above-set external environment parameters of the container, the spatial position and size of the items inside the container, the state parameters of the items inside the container and the corresponding environmental parameter values ​​in all grids; the simulation data includes temperature, pressure and humidity; the environmental parameter values ​​include temperature, pressure and humidity;

[0007] S2: training the neural network model of the environment inside the container: using the data stored in S1, a deep neural network is used to establish multiple different neural network models for the spatial position and size of the items inside the container, wherein different neural network models correspond to different spatial positions and sizes of the items inside the container;

[0008] S3: Estimate the actual spatial location and size of items inside the container;

[0009] S4: Matching neural network model: According to the actual spatial position and size of the items inside the container obtained in S3, the corresponding neural network model is searched in S2, represented by NN;

[0010] S5: Obtaining the state parameters of the items inside the container: Inputting the position of the environmental sensor placed in the actual container and the measured values ​​of the environmental parameters at different times at the position, and the external environmental parameter values ​​of the container determined by the weather forecast into the NN model to obtain the state parameters of the items inside the container; repeating steps S3-S5 until all the items inside the actual container are traversed;

[0011] S6: Predicting environmental parameters inside the container: Based on the internal item status parameters of the container obtained in S5 and the external environmental parameter values ​​of the container determined by the weather forecast, the NN model is used to predict the temperature, pressure and humidity at any position in the container.

[0012] Preferably, setting the spatial position and size of items inside the container in S1 includes: setting a minimum cuboid for accommodating the items inside the container, the grid coordinates of the six vertices of the minimum cuboid being represented by the grid coordinate system in S1, and the grid coordinates of the six vertices representing the spatial position and size of the items inside the container.

[0013] Preferably, estimating the actual spatial position and size of items inside the container in S3 includes: estimating the actual volume of items inside the container according to the mass and density of the items inside the container, setting a minimum cuboid for accommodating the items inside the container, the grid coordinates of the six vertices of the minimum cuboid being represented by the grid coordinate system in S1, and the grid coordinates of the six vertices representing the actual spatial position and size of items inside the container.

[0014] The S1 comprises the following steps:

[0015] S11: Divide the internal space of the container into three-dimensional grids, divide it into grids of the same size, and establish a grid coordinate system;

[0016] S12: Setting container external environment parameters;

[0017] S13: Setting the spatial position and size of the items inside the container: setting a minimum cuboid for accommodating the items inside the container, wherein the grid coordinates of the six vertices of the minimum cuboid are represented by the grid coordinate system, and the grid coordinates of the six vertices represent the spatial position and size of the items inside the container;

[0018] S14: Setting the status parameters of items inside the container;

[0019] S15: Perform finite element simulation on the internal environment temperature, pressure and humidity of the container to obtain environmental parameter values ​​in all grids; the environmental parameter values ​​in the grids include temperature, pressure and humidity;

[0020] S16: Store the external environment parameters of the container set in S12, the spatial position and size of the items inside the container set in S13, the state parameters of the items inside the container set in S14, and the environmental parameter values ​​in all grids obtained corresponding to the above parameters; store the temperature, pressure and humidity in all grids obtained in S15;

[0021] S17: Return to S14, change the state parameters of the items inside the container, repeat steps S15-S17, and obtain the environmental parameter values ​​in all grids;

[0022] S18: Return to S13, change the spatial position and size of the items inside the container, repeat steps S14-S17, and obtain the environmental parameter values ​​in all grids;

[0023] S19: Return to S12, change the external environmental parameters of the container, repeat steps S13-S17, and obtain the environmental parameter values ​​in all grids.

[0024] Preferably, the external environmental parameters in S12 include temperature, light, air flow rate, air humidity, air pressure, and heat exchange coefficient.

[0025] Preferably, the state parameters of items inside the container in S14 include heat absorption rate, specific heat capacity, water absorption rate, and air permeability.

[0026] Preferably, the neural network model in S2 represents the correspondence between the environmental parameters in different grids of the container and the external environmental parameters, the state parameters of the items inside the container and the grid positions.

[0027] Preferably, the grid position is obtained through a grid coordinate system established by S1.

[0028] The neural network model described in S2 satisfies the following conditions:

[0029]

[0030] Among them, t i,j,k , p i,j,k ,h i,j,k The temperature, pressure and humidity of the space inside the container obtained by the finite element simulation in step S1; The estimated values ​​of temperature, pressure and humidity inside the container obtained by the deep neural network; n l , nw , n h Respectively refer to the number of grids in the length parallel direction, width parallel direction and height parallel direction inside the container.

[0031] Beneficial effects:

[0032] The present invention provides a method for quickly estimating container environmental parameters, which performs finite element simulation on the environment inside the container, obtains simulation data, and ensures the rationality and reliability of the simulation data; divides the internal space of the container into three-dimensional grids, establishes a grid coordinate system, and is convenient for obtaining the spatial position and size of items inside the container; simulates and sets the external environmental parameters of the container, the spatial position and size of items inside the container, and the state parameters of items inside the container, and performs finite element simulation on the internal environmental temperature, pressure, and humidity of the container to obtain the environmental parameter values ​​of all grids, further ensuring the rationality and reliability of the simulation data, thereby ensuring the accuracy of the prediction of the subsequently established model; the simulated external environmental parameters of the container, the spatial position and size of the items inside the container, and the state parameters of the items inside the container are set. The spatial position and size of the items, the state parameters of the items inside the container, and the corresponding environmental parameter values ​​in all grids are stored for establishing a neural network model; the neural network model can obtain the state parameters of the items inside the container according to the position of the environmental sensor placed in the actual container and the environmental parameter measurement values ​​of the position at different times and the external environmental parameters of the container determined by the weather forecast; the above-mentioned state parameters of the items inside the container are combined with the external environmental parameters of the container determined by the weather forecast and input into the neural network model to predict the temperature, pressure and humidity at any position in the container, solving the problem that the prior art only installs the perception sensor in one place in the container and cannot directly obtain the overall environmental state of the container. The present invention uses finite element simulation to obtain simulation data to establish a neural network model of environmental parameters inside the container, uses online data of the location of the sensor in the container to estimate the temperature, pressure and humidity of the items in the container, and on this basis uses a neural network model to quickly obtain the temperature, pressure and humidity of each position in the container, so as to facilitate monitoring of the temperature, pressure and humidity inside the container. BRIEF DESCRIPTION OF THE DRAWINGS

[0033] Figure 1 Flow chart of the scheme of the present invention.

[0034] Figure 2 It is a flow chart of step S1 of the embodiment of the present invention. DETAILED DESCRIPTION

[0035] The present invention will be further described below in conjunction with the accompanying drawings:

[0036] A fast estimation method for container environmental parameters, such as Figure 1 As shown, it includes the following steps:

[0037] S1: Perform finite element simulation on the environment inside the container and obtain simulation data: divide the internal space of the container into three-dimensional grids and establish a grid coordinate system; set the external environment parameters of the container; set the spatial position and size of the items inside the container: set a minimum cuboid for accommodating the internal items for the items inside the container, and the grid coordinates of the six vertices of the minimum cuboid are expressed by the grid coordinate system as (ls1, ws1, hs1), (ls2, ws2, hs2), (ls3, ws4, hs4), (ls5, ws5, hs5), (ls6, ws6, hs6), and the grid coordinates of the six vertices represent the spatial position and size of the internal items; set the state parameters of the items inside the container; perform finite element simulation on the temperature, pressure and humidity of the internal environment of the container to obtain the environmental parameter values ​​in all grids; store the above-set external environment parameters of the container, the spatial position and size of the internal items of the container, the state parameters of the internal items of the container, and the corresponding environmental parameter values ​​in all grids; the simulation data include temperature, pressure and humidity; the environmental parameter values ​​include temperature, pressure and humidity; Figure 2 As shown, the following steps are included:

[0038] S11: Divide the internal space of the container into three-dimensional grids, divide it into grids of the same size, and establish a grid coordinate system; in a specific embodiment of the present invention, the interval of each side parallel to the length of the container is set to ls, the interval of each side parallel to the width of the container is set to ws, and the interval of each side parallel to the height of the container is set to hs, then the space size of each grid obtained by division is ls×ws×hs;

[0039] S12: Setting the external environment parameters of the container, wherein the external environment parameters include temperature, light, air velocity, air humidity, air pressure, and heat exchange coefficient; the external environment parameters of the container are set by meta-simulation;

[0040] S13: Setting the spatial position and size of the items inside the container: a minimum cuboid is set for the items inside the container to accommodate the items inside the container, and the grid coordinates of the six vertices of the minimum cuboid are represented by the grid coordinate system in S11 as (ls1, ws1, hs1), (ls2, ws2, hs2), (ls3, ws4, hs4), (ls5, ws5, hs5), (ls6, ws6, hs6), and the grid coordinates of the six vertices represent the spatial position and size of the items inside the container; the spatial position and size of the items inside the container are set by meta-simulation, and only one item inside the container is set during meta-simulation;

[0041] S14: setting state parameters of items inside the container, wherein the state parameters of items inside the container include heat absorption rate, specific heat capacity, water absorption rate, and air permeability; the state parameters of items inside the container are set by meta-simulation;

[0042] S15: Perform finite element simulation on the internal environment temperature, pressure and humidity of the container to obtain the environmental parameter values ​​in all grids; the environmental parameter values ​​in the grids include temperature, pressure and humidity; the environmental parameter value of the i-th grid space can be expressed as (ti, pi, hi);

[0043] S16: Store the external environment parameters of the container set in S12, the space range of the cuboid occupied by the items inside the container set in S13 (the coordinates of the six vertices of the cuboid), the state parameters of the items inside the container set in S14, and the environment parameter values ​​in all grids obtained in S15 corresponding to the above parameters;

[0044] S17: Return to S14, change the state parameters of the items inside the container, repeat steps S15-S17, obtain the environmental parameter values ​​in all grids, and traverse all the state parameters of the items inside the working range of the container; in a specific embodiment of the present invention, each time returning to S14, only one state parameter of one of the items inside the container is changed.

[0045] S18: Return to S13, change the position and size of the space of the items inside the container, repeat steps S14-S17, obtain the environmental parameter values ​​in all grids, until the rectangular space occupied by the items inside the container is traversed, that is, it is necessary to traverse all possible positions and sizes of the rectangular space occupied by the items inside the container (six vertex coordinates);

[0046] S19: Return to S12, change the external environmental parameters of the container, repeat steps S13-S17, obtain the environmental parameter values ​​in all grids, until all external environmental parameters within the working range of the container are traversed; in a specific embodiment of the present invention, in order to facilitate decoupling and classification, each time S12 is returned, only one of the external environmental parameters of the container is changed; wherein, the simulation data includes temperature, pressure and humidity; the environmental parameter values ​​include temperature, pressure and humidity.

[0047] S2: Training the neural network model of the internal environment of the container: using the data stored in S1, a deep neural network is used to establish a neural network model of the relationship between the internal environment parameters of multiple containers in different grids and the external environment parameters, the state parameters of the internal items of the container, and the grid position, wherein different internal space positions and sizes of the items in the container correspond to different neural network models;

[0048] The grid position is obtained by establishing a grid coordinate system established by S1; the neural network model satisfies the following conditions:

[0049]

[0050] Among them, t i,j,k , p i,j,k ,h i,j,k The temperature, pressure and humidity of the space inside the container obtained by the finite element simulation in step S1; The estimated values ​​of temperature, pressure and humidity inside the container obtained by the deep neural network; n l , n w , n h Respectively refer to the number of grids in the length parallel direction, width parallel direction and height parallel direction inside the container.

[0051] S3: Estimate the spatial position and size of items inside the actual container: When the actual container transports goods, the mass and density of the items in the container are known, and the volume of the object can be estimated by the mass and density of the items. Then, a minimum cuboid is set to accommodate the items inside the actual container. The grid coordinates of the six vertices of the minimum cuboid are represented by the grid coordinate system described in S1. The grid coordinates of the six vertices represent the spatial position and size of the items inside the actual container.

[0052] S4: Matching neural network model: According to the actual spatial position and size of the items in the container obtained in S3, the corresponding neural network model is searched in S2, represented by NN; several neural network models are established in S2, where different neural network models correspond to different spatial positions and sizes of items in the container. The actual spatial position and size of the items in the container obtained in step S3 are used to search several neural network models in S2. If the actual spatial position and size of the items in the container are: (ls1 a ,ws1 a ,hs1 a ), (ls2 a ,ws2 a ,hs2 a ), (ls3 a ,ws4 a ,hs4 a ), (ls5 a ,ws5 a ,hs5 a ), (hs6 a ,ws6 a ,hs6 a ) The spatial position and size of the items inside the container corresponding to the neural network model in S2 (assuming that: (ls1 b ,ws1b ,hs1 b ), (ls2 b ,ws2 b ,hs2 b ), (ls3 b ,ws4 b ,hs4 b ), (ls5 b ,ws5 b ,hs5 b ), (ls6 b ,ws6 b ,hs6 b )) is the same as: ls1 a =ls1 b 、ws1 a =ws1 b 、hs1 a =hs1 b .......hs6 a =hs6 b 、ws6 a =ws6 b 、hs6=hs6 b , then we get the matching neural network model, represented by NN.

[0053] S5: Obtaining the state parameters of the items inside the container: Input the position of the environmental sensor placed in the actual container, the environmental parameter measurement values ​​of the position at different times, and the external environmental parameter values ​​of the container determined by the weather forecast into the NN model to obtain the state parameters of the items inside the container; repeat steps S3-S5 until all the items inside the container are traversed.

[0054] S6: Predicting environmental parameters inside the container: Based on the state parameters of the items inside the container obtained in S5 and the external environmental parameter values ​​of the container determined by the weather forecast, the NN model can be used to predict the temperature, pressure and humidity at any position in the container.

Claims

1. A method for quickly estimating container environmental parameters, characterized in that: The steps include: S1: Perform finite element simulation on the environment inside the container to obtain simulation data: divide the internal space of the container into three-dimensional grids and establish a grid coordinate system; set the external environment parameters of the container; set the spatial position and size of the items inside the container; set the state parameters of the items inside the container; perform finite element simulation on the temperature, pressure and humidity of the internal environment of the container to obtain the environmental parameter values ​​in all grids; store the above-set external environment parameters of the container, the spatial position and size of the items inside the container, the state parameters of the items inside the container and the corresponding environmental parameter values ​​in all grids; the simulation data includes temperature, pressure and humidity; the environmental parameter values ​​include temperature, pressure and humidity; the steps include: S11: Divide the internal space of the container into three-dimensional grids, divide it into grids of the same size, and establish a grid coordinate system; S12: Setting container external environment parameters; S13: Setting the spatial position and size of the items inside the container: setting a minimum cuboid for accommodating the items inside the container, wherein the grid coordinates of the six vertices of the minimum cuboid are represented by the grid coordinate system, and the grid coordinates of the six vertices represent the spatial position and size of the items inside the container; S14: Setting the status parameters of items inside the container; S15: Perform finite element simulation on the internal environment temperature, pressure and humidity of the container to obtain environmental parameter values ​​in all grids; the environmental parameter values ​​in the grids include temperature, pressure and humidity; S16: Store the external environment parameters of the container set in S12, the spatial position and size of the items inside the container set in S13, the state parameters of the items inside the container set in S14, and the environmental parameter values ​​in all grids obtained corresponding to the above parameters; store the temperature, pressure and humidity in all grids obtained in S15; S17: Return to S14, change the state parameters of the items inside the container, repeat steps S15-S17, and obtain the environmental parameter values ​​in all grids; S18: Return to S13, change the spatial position and size of the items inside the container, repeat steps S14-S17, and obtain the environmental parameter values ​​in all grids; S19: Return to S12, change the external environmental parameters of the container, repeat steps S13-S17, and obtain the environmental parameter values ​​in all grids; S2: Training the neural network model of the container internal environment: Using the data stored in S1, a deep neural network is used to establish multiple different neural network models for the spatial position and size of the items inside the container, wherein different neural network models correspond to different spatial positions and sizes of the items inside the container; the neural network model meets the following conditions: Among them, t i,j,k , p i,j,k ,h i,j,k The temperature, pressure and humidity of the space inside the container obtained by the finite element simulation in step S1; The estimated values ​​of temperature, pressure and humidity inside the container obtained by the deep neural network; n l , n w , n h Respectively refer to the number of grids in the length parallel direction, width parallel direction and height parallel direction in the container; S3: Estimate the actual spatial location and size of items inside the container; S4: Matching neural network model: According to the actual spatial position and size of the items inside the container obtained in S3, the corresponding neural network model is searched in S2, represented by NN; S5: Obtaining the state parameters of the items inside the container: Inputting the position of the environmental sensor placed in the actual container and the measured values ​​of the environmental parameters at different times at the position, and the external environmental parameter values ​​of the container determined by the weather forecast into the NN model to obtain the state parameters of the items inside the container; repeating steps S3-S5 until all the items inside the actual container are traversed; S6: Predicting environmental parameters inside the container: Based on the state parameters of the items inside the container obtained in S5 and the external environmental parameter values ​​of the container determined by the weather forecast, the NN model is used to predict the temperature, pressure and humidity at any position in the container.

2. The method for rapid estimation of container environmental parameters according to claim 1, characterized in that: The spatial position and size of items inside the container are set in S1, including: setting a minimum cuboid for accommodating the items inside the container, the grid coordinates of the six vertices of the minimum cuboid are represented by the grid coordinate system in S1, and the grid coordinates of the six vertices represent the spatial position and size of the items inside the container.

3. The method for rapid estimation of container environmental parameters according to claim 1, characterized in that: The actual spatial position and size of the items inside the container are estimated in S3, including: estimating the volume of the items inside the container according to the mass and density of the items inside the container, setting a minimum cuboid for accommodating the items inside the container, and the grid coordinates of the six vertices of the minimum cuboid are represented by the grid coordinate system in S1, and the grid coordinates of the six vertices represent the actual spatial position and size of the items inside the container.

4. The method for rapidly estimating container environmental parameters according to any one of claims 1 to 3, characterized in that: The external environmental parameters described in S12 include temperature, light, air flow rate, air humidity, air pressure, and heat exchange coefficient.

5. The method for quickly estimating container environmental parameters according to any one of claims 1 to 3, characterized in that: The state parameters of the items inside the container described in S14 include heat absorption rate, specific heat capacity, water absorption rate, and air permeability.

6. The method for rapid estimation of container environmental parameters according to claim 1, characterized in that: The neural network model described in S2 represents the correspondence between the environmental parameters in different grids of the container and the external environmental parameters, the state parameters of the items inside the container and the grid positions.

7. The method for rapid estimation of container environmental parameters according to claim 6, characterized in that: The grid position is obtained through the grid coordinate system established by S1.

Images