Carrying equipment credible positioning method and positioning device thereof

Abstract

The invention discloses a credible positioning method and device for carrying equipment. The carrying equipment is at least provided with a satellite positioning device, a Bluetooth positioning device and an environment sensor. The positioning method comprises the following steps: acquiring satellite positioning data acquired by a satellite positioning device, Bluetooth positioning data acquired by a Bluetooth positioning device and environment data acquired by an environment sensor; inputting the satellite positioning data, the Bluetooth positioning data and the environment data into a positioning model obtained by training a deep learning model based on a data sample to obtain a first predicted position of the carrying equipment; calculating a second predicted position of the carrying equipment based on the satellite positioning data and the Bluetooth positioning data; determining a first weight coefficient and a second weight coefficient based on a prediction error index of the positioning model; and determining a final predicted position of the carrying equipment based on the first predicted position, the first weight coefficient, the second predicted position and the second weight coefficient. And the positioning precision and reliability of the carrying equipment are improved.

Description

Technical Field

[0001] The present application relates to the field of positioning technology, and more particularly to a method for reliable positioning of a transportation equipment and its positioning device. Background Art

[0002] At present, during the transportation of transportation equipment such as, but not limited to, tank trucks for hazardous chemicals (hereinafter referred to as hazardous chemicals), accurate knowledge of its location is crucial for safety management. A Beidou positioning system and a Bluetooth positioning system are usually set on the transportation equipment. The Beidou positioning system can provide positioning services in a relatively wide outdoor area, while Bluetooth positioning has its unique advantages in specific short-distance or indoor environments. The existing technology calculates the predicted position of the transportation equipment by using a fusion algorithm based on the Beidou positioning data collected by the Beidou positioning system and the Bluetooth positioning data collected by the Bluetooth positioning system. However, the traditional fusion positioning method still has deficiencies in terms of accuracy, low power consumption, and adaptability to complex environments. Summary of the Invention

[0003] The present application is proposed to solve the above problems. According to one aspect of the present application, a method for reliable positioning of a transportation equipment is provided. At least a satellite locator, a Bluetooth locator, and an environmental sensor are set on the transportation equipment. The method for reliable positioning of the transportation equipment includes:

[0004] Obtaining satellite positioning data collected by the satellite locator, Bluetooth positioning data collected by the Bluetooth locator, and environmental data collected by the environmental sensor;

[0005] Inputting the satellite positioning data, the Bluetooth positioning data, and the environmental data into a positioning model to obtain a first predicted position of the transportation equipment; wherein, the positioning model is obtained by training a deep learning model based on data samples, and the data samples include: satellite positioning data samples collected based on the satellite locator, Bluetooth positioning data samples collected based on the Bluetooth locator, and environmental data samples collected based on the environmental sensor;

[0006] Calculating a second predicted position of the transportation equipment based on the satellite positioning data and the Bluetooth positioning data;

[0007] Determining a first weight coefficient corresponding to the first predicted position and a second weight coefficient corresponding to the second predicted position based on the prediction error index of the positioning model;

[0008] Determining the final predicted position of the transportation equipment based on the first predicted position, the first weight coefficient, the second predicted position, and the second weight coefficient.

[0009] In some embodiments of the present application, the sum of the first weight coefficient and the second weight coefficient is equal to 1;

[0010] Determine the final predicted position of the carrying equipment based on the first predicted position, the first weight coefficient, the second predicted position, and the second weight coefficient, including:

[0011] Take the sum of the product of the first longitude in the first predicted position and the first weight coefficient and the product of the second longitude in the second predicted position and the second weight coefficient as the final predicted longitude in the final predicted position;

[0012] Take the sum of the product of the first latitude in the first predicted position and the first weight coefficient and the product of the second latitude in the second predicted position and the second weight coefficient as the final predicted latitude in the final predicted position.

[0013] In some embodiments of the present application, the reliable positioning method for the carrying equipment further includes:

[0014] When the environmental sensor detects that the carrying equipment is in a stationary state and the change range of the first predicted position obtained based on the positioning model within a set time period does not exceed the set range, reduce the sampling frequency of the satellite locator and the Bluetooth locator from the first sampling frequency to the second sampling frequency, or control the satellite locator and the Bluetooth locator to switch to the sleep mode;

[0015] When the environmental sensor detects that the carrying equipment starts to move again, restore the sampling frequency of the satellite locator and the Bluetooth locator to the first sampling frequency.

[0016] In some embodiments of the present application, both the satellite positioning data and the satellite positioning data samples include at least one of satellite signal strength, satellite elevation angle, and the number of satellites; and / or,

[0017] Both the Bluetooth positioning data and the Bluetooth positioning data samples include at least one of Bluetooth signal strength, Bluetooth signal arrival angle, and the connection status between the Bluetooth locator and surrounding Bluetooth beacons; and / or,

[0018] The environmental sensor includes at least one of an accelerometer, a gyroscope, a barometer, and a camera device, and both the environmental data and the environmental data samples include at least one of the acceleration of the carrying equipment, the angular velocity of the carrying equipment, the detected barometric pressure change of the barometer, and the indoor / outdoor identification of the environment around the carrying equipment.

[0019] In some embodiments of the present application, the reliable positioning method for the carrying equipment further includes:

[0020] Collect satellite positioning data samples, Bluetooth positioning data samples, and environmental data samples of the carrying equipment during operation in different transportation environments at a set sampling frequency to form raw data samples arranged in a time series;

[0021] Clean the raw data samples;

[0022] Calculate the feature vectors of the original data samples at each time point after cleaning, and form the feature vectors at multiple time points into a three-dimensional data tensor with a shape of (time series length, feature vector dimension); wherein, different types of data in the original data samples are arranged in the feature vectors in sequence;

[0023] According to the reference positioning system, label the true position information corresponding to the original data samples at each time point;

[0024] Perform normalization processing on the feature vectors in the three-dimensional data tensor, and use the normalized three-dimensional data tensor as the data sample.

[0025] In some embodiments of the present application, the method for reliable positioning of a carrying device further includes:

[0026] Divide the data samples into a training set, a validation set, and a test set;

[0027] Use the training set to perform multiple rounds of training on the deep learning model, and after each round of training, use the validation set to evaluate the loss value of the deep learning model;

[0028] Stop training when the loss value no longer decreases or starts to increase for a continuously set number of rounds, and use the deep learning model obtained after training as the positioning model;

[0029] Use the test set to evaluate the prediction error index of the positioning model.

[0030] In some embodiments of the present application, the deep learning model includes: a hybrid architecture composed of a convolutional neural network and a long short-term memory network;

[0031] Among them, the input format of the input layer of the hybrid architecture is: a three-dimensional data tensor with a shape of (time series length, feature vector dimension); and / or,

[0032] The convolutional layer of the hybrid architecture is multiple convolutional kernels with a size of (3, feature vector dimension), and the convolution step of the convolutional layer is set to a set step; and / or,

[0033] The padding method of the hybrid architecture is the SAME padding method; and / or,

[0034] The activation function of the hybrid architecture is a linear activation function; and / or,

[0035] The maximum layer pooling window size of the hybrid architecture is (2, 1), and the long short-term memory network sets multiple hidden units; and / or,

[0036] The number of neurons in the fully connected layer of the hybrid architecture is determined based on the positioning accuracy requirements; and / or,

[0037] Define the loss function of the hybrid architecture as the mean square error function, and define the optimizer of the hybrid architecture as the Adam optimizer.

[0038] According to another aspect of the present application, there is provided a reliable positioning device for a transportation equipment. At least a satellite locator, a Bluetooth locator and an environmental sensor are provided on the transportation equipment. The reliable positioning device for the transportation equipment includes: a storage medium and a processor. A computer program is stored on the storage medium and run by the processor. When the computer program is run by the processor, the processor executes any one of the above-mentioned reliable positioning methods for the transportation equipment.

[0039] According to the reliable positioning method and positioning device for the transportation equipment provided by the embodiments of the present application, a positioning model is obtained by pre-training a deep learning model with data samples including satellite positioning data samples, Bluetooth positioning data samples and environmental data samples. During the positioning process of the transportation equipment, the real-time obtained satellite positioning data, Bluetooth positioning data and environmental data are input into the pre-trained positioning model to obtain the first predicted position of the transportation equipment, and the second predicted position of the transportation equipment is calculated based on the satellite positioning data and the Bluetooth positioning data. The first weight coefficient corresponding to the first predicted position and the second weight coefficient corresponding to the second predicted position are determined based on the prediction error index of the positioning model, and then the final predicted position of the transportation equipment is determined based on the first predicted position, the first weight coefficient, the second predicted position and the second weight coefficient, thereby improving the positioning accuracy and reliability of the transportation equipment and the credibility of the final predicted position. BRIEF DESCRIPTION OF THE DRAWINGS

[0040] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the following will briefly introduce the drawings required for the description of the embodiments. Obviously, the following drawings are only some embodiments of the present application. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0041] Figure 1 It is a schematic flow chart of a reliable positioning method for a transportation equipment shown in an embodiment of the present application;

[0042] Figure 2 It is a flow chart of determining the final predicted position based on the first predicted position and the second predicted position shown in an embodiment of the present application;

[0043] Figure 3 It is a simplified schematic flow chart of a reliable positioning method for a transportation equipment shown in an embodiment of the present application. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0044] To make the objectives, technical solutions, and advantages of the present application more apparent, the following will describe exemplary embodiments according to the present application in detail with reference to the accompanying drawings. Apparently, the described embodiments are only a part of the embodiments of the present application, rather than all embodiments of the present application. It should be understood that the present application is not limited by the exemplary embodiments described herein. Based on the embodiments of the present application described herein, all other embodiments obtained by those skilled in the art without creative efforts should fall within the protection scope of the present application.

[0045] In the following description, numerous specific details are given to provide a more thorough understanding of the present application. However, it is obvious to those skilled in the art that the present application can be implemented without one or more of these details. In other instances, in order to avoid confusion with the present application, some well-known technical features are not described.

[0046] It should be understood that the present application can be implemented in different forms and should not be construed as limited to the embodiments presented herein. On the contrary, providing these embodiments will make the disclosure thorough and complete, and will fully convey the scope of the present application to those skilled in the art.

[0047] The purpose of the terms used herein is only to describe specific embodiments and is not a limitation of the present application. When used herein, the singular forms "a", "an", and "the" are also intended to include the plural forms unless the context clearly indicates otherwise. It should also be understood that the terms "comprising" and / or "including", when used in this specification, identify the presence of the described features, integers, steps, operations, elements, and / or components, but do not preclude the presence or addition of one or more other features, integers, steps, operations, elements, components, and / or groups. When used herein, the term "and / or" includes any and all combinations of the related listed items.

[0048] To thoroughly understand the present application, detailed structures will be presented in the following description to illustrate the technical solutions proposed by the present application. The optional embodiments of the present application are described in detail below. However, in addition to these detailed descriptions, the present application can also have other implementation manners.

[0049] To solve at least some of the above-mentioned related technical problems, the following embodiments are proposed in the present application.

[0050] The following will describe in detail some embodiments of the present application with reference to the accompanying drawings. Without conflict, the following embodiments and the features in the embodiments can be combined with each other.

[0051] Embodiment 1

[0052] First, introduce the application scenario of the positioning method of the carrying equipment shown in the example of this application. The positioning method of the carrying equipment is applied in the positioning process of the carrying equipment to predict the current position of the carrying equipment. Among them, at least a satellite locator, a Bluetooth locator, and an environmental sensor are provided on the carrying equipment. The carrying equipment can be various types of carrying equipment. Exemplarily, the carrying equipment can be a carrying equipment of a vehicle type such as, but not limited to, a hazardous chemical tanker. Of course, the carrying equipment is not limited to the carrying equipment of the vehicle type. For example, it can also be a container-type carrying equipment.

[0053] Exemplarily, referring to Figure 1 , the reliable positioning method of the carrying equipment provided by the embodiment of this application mainly includes the following steps:

[0054] In S101, obtain the satellite positioning data collected by the satellite locator, the Bluetooth positioning data collected by the Bluetooth locator, and the environmental data collected by the environmental sensor;

[0055] In S102, input the satellite positioning data, the Bluetooth positioning data, and the environmental data into the positioning model to obtain the first predicted position of the carrying equipment; among them, the positioning model is obtained by training a deep learning model based on data samples, and the data samples include: satellite positioning data samples collected based on the satellite locator, Bluetooth positioning data samples collected based on the Bluetooth locator, and environmental data samples collected based on the environmental sensor;

[0056] In S103, calculate the second predicted position of the carrying equipment based on the satellite positioning data and the Bluetooth positioning data;

[0057] In S104, based on the prediction error index of the positioning model, determine the first weight coefficient corresponding to the first predicted position and the second weight coefficient corresponding to the second predicted position;

[0058] In S105, determine the final predicted position of the carrying equipment based on the first predicted position, the first weight coefficient, the second predicted position, and the second weight coefficient.

[0059] In the above solution, a positioning model is obtained by pre-training a deep learning model with data samples including satellite positioning data samples, Bluetooth positioning data samples, and environmental data samples. During the positioning process of the transportation equipment, the real-time obtained satellite positioning data, Bluetooth positioning data, and environmental data are input into the pre-trained positioning model to obtain the first predicted position of the transportation equipment, and the second predicted position of the transportation equipment is calculated based on the satellite positioning data and the Bluetooth positioning data. The first weight coefficient corresponding to the first predicted position and the second weight coefficient corresponding to the second predicted position are determined based on the prediction error index of the positioning model. Then, the final predicted position of the transportation equipment is determined based on the first predicted position, the first weight coefficient, the second predicted position, and the second weight coefficient. Based on the great advantages demonstrated by deep learning technology in processing multi-source data fusion and complex pattern recognition, when determining the first predicted position, not only the satellite positioning data and the Bluetooth positioning data are utilized, but also the environmental data collected by the environmental sensors of the transportation equipment is utilized to adapt to complex transportation scenarios, so that the first predicted position output by the trained positioning model is closer to the real position. Subsequently, the final predicted position is jointly obtained based on the first predicted position and the second predicted position, so as to be able to assist in correcting the second predicted position calculated by using the traditional fusion algorithm, thereby narrowing the difference between the final predicted position and the real position, improving the positioning accuracy and reliability of the transportation equipment, enhancing the credibility of the final predicted position, and providing strong technical support for, such as but not limited to, the safety supervision of hazardous chemical transportation.

[0060] The above steps will be introduced in detail below with reference to the accompanying drawings.

[0061] First, refer to Figure 1 to obtain the satellite positioning data collected by the satellite positioning device, the Bluetooth positioning data collected by the Bluetooth positioning device, and the environmental data collected by the environmental sensor.

[0062] Exemplarily, a data acquisition unit can be set up to be communicatively connected to the satellite positioning device, the Bluetooth positioning device, and the environmental sensor. The data acquisition unit collects the satellite positioning data, Bluetooth positioning data, and environmental data newly collected by the satellite positioning device, the Bluetooth positioning device, and the environmental sensor at each set time interval. For example, the satellite positioning device collects satellite positioning data every 1 second, the Bluetooth positioning device collects Bluetooth positioning data every 1 second, the environmental sensor collects environmental data every 1 second, and the data acquisition unit collects the satellite positioning data, Bluetooth positioning data, and environmental data newly collected by the satellite positioning device, the Bluetooth positioning device, and the environmental sensor every 1 second.

[0063] Next, refer to Figure 1, the satellite positioning data, Bluetooth positioning data, and environmental data are input into a positioning model obtained by pre-training a deep learning model based on data samples, and the first predicted position of the carrier equipment is obtained. That is, the types of input data for this positioning model are three categories, namely satellite positioning data collected by a satellite positioner, Bluetooth positioning data collected by a Bluetooth positioner, and environmental data collected by an environmental sensor. And the output of this positioning model is the first predicted position of the carrier equipment. Thus, the first predicted position of the carrier equipment can be obtained based on these three categories of data: satellite positioning data, Bluetooth positioning data, and environmental data.

[0064] In the process of training a deep learning model based on data samples to obtain a positioning model, various methods can be adopted. Some methods are introduced exemplarily as follows.

[0065] Exemplarily, referring to Figure 3 , this reliable positioning method for the carrier equipment may further include: pre-constructing data samples including satellite positioning data samples, Bluetooth positioning data samples, and environmental data samples; constructing a deep learning model, and training the deep learning model based on the data samples to obtain a positioning model. Since the data samples used in training the deep learning model include these three categories of data: satellite positioning data samples, Bluetooth positioning data samples, and environmental data samples, the obtained positioning model can be adapted to complex and changeable transportation environments, so as to improve the positioning accuracy and reliability of the carrier equipment.

[0066] Regarding the types of the above satellite positioning data and satellite positioning data samples, various types of data can be included. Exemplarily, the satellite positioning data and satellite positioning data samples may both include at least one of satellite signal strength, satellite elevation angle, and number of satellites.

[0067] Regarding the types of satellite positioners, they can be various types of satellite positioning devices. Exemplarily, the satellite positioner can be, such as but not limited to, a Beidou positioner, a GPS positioner, etc.

[0068] Regarding the types of Bluetooth positioning data and Bluetooth positioning data samples, various types of data can be included. Exemplarily, the Bluetooth positioning data and Bluetooth positioning data samples may both include at least one of Bluetooth signal strength, Bluetooth signal arrival angle, and connection status between the Bluetooth positioner and surrounding Bluetooth beacons.

[0069] Regarding the types of environmental sensors, they can be various types of sensors installed on the carrier equipment. For example, the environmental sensor may include at least one of an accelerometer, a gyroscope, a barometer, and a camera device.

[0070] Regarding the types of environmental data and environmental data samples, they can be data that can characterize the environment where the transportation equipment is located, the operating state of the transportation equipment, etc., detected by any type of sensor on the transportation equipment. For example, the environmental data and environmental data samples can both include at least one of the following: the acceleration of the transportation equipment, the angular velocity of the transportation equipment, the pressure change detected by the barometer, and the indoor / outdoor identification of the surrounding environment of the transportation equipment.

[0071] There are various ways to pre-construct data samples including satellite positioning data samples, Bluetooth positioning data samples, and environmental data samples, that is, to perform multi-source data collection. Exemplarily, referring to Figure 3 , the reliable positioning method for transportation equipment can further include the following steps:

[0072] Pre-collect satellite positioning data samples, Bluetooth positioning data samples, and environmental data samples of the transportation equipment during operation in different transportation environments at a set sampling frequency to form raw data samples arranged in a time series;

[0073] Clean the raw data samples;

[0074] Calculate the feature vectors of the raw data samples at each time point after cleaning, and form a three-dimensional data tensor with the shape of (time series length, feature vector dimension) by combining the feature vectors at multiple time points; among them, different types of data in the raw data samples are arranged in the feature vectors in sequence;

[0075] According to the reference positioning system, label the true position information corresponding to the raw data samples at each time point;

[0076] Normalize the feature vectors in the three-dimensional data tensor, and use the normalized three-dimensional data tensor as the data sample.

[0077] By cleaning, labeling, and normalizing the raw data samples arranged in a time series, it is possible to remove obvious abnormal data points, organize the cleaned data into a format suitable for input to the deep learning model according to the time series, label the true position information for each raw data sample, and perform normalization processing to map the data to the [0,1] interval, so as to facilitate the training of the deep learning model.

[0078] Exemplarily, there are various ways to collect satellite positioning data samples, Bluetooth positioning data samples, and environmental data samples of a carrier equipment during operation in different transportation environments in advance at a set sampling frequency to form original data samples arranged in a time series. For example, in practical applications, first install a Beidou positioning device, a Bluetooth positioning device, and an environmental sensor on a carrier equipment such as, but not limited to, a hazardous chemical tanker, and connect the Beidou positioning device, the Bluetooth positioning device, and the environmental sensor to a data acquisition unit for communication. According to the steps of data collection and preprocessing, let the carrier equipment continuously collect data samples at a certain time interval (such as every 1 second) in different transportation environments (including urban roads, highways, warehouses, etc.), and the collection time can last for several weeks to obtain rich and diverse data samples and get original data samples arranged in a time series. Then, perform cleaning, formatting, and annotation processing on the collected original data samples to facilitate the construction of a deep learning model and its training and evaluation.

[0079] Exemplarily, referring to Figure 3 , the above cleaning, annotation, and normalization processing can be data preprocessing steps.

[0080] Exemplarily, there are various ways to clean the original data samples. For example, eliminate data with satellite signal strength exceeding a reasonable range, where the reasonable range can be a range determined based on historical data and satellite signal propagation theory. For example, eliminate data with suddenly large fluctuations in Bluetooth signal strength and not conforming to the Bluetooth signal propagation law to remove obvious abnormal data points.

[0081] Exemplarily, the cleaned data can be sorted into a format suitable for input to a deep learning model according to the time series, and the original data sample at each time point is regarded as a feature vector, and different types of data are arranged in the feature vector in sequence. For example, the types of different types of data arranged in the feature vector can be in the following way: [Beidou satellite signal strength 1, Beidou satellite elevation angle 1, Bluetooth signal strength 1, Bluetooth arrival angle 1, carrier equipment acceleration 1,...]. Feature vectors at multiple time points form a three-dimensional data tensor with a shape of (time series length, feature vector dimension).

[0082] Exemplarily, there are various ways to label the true position information corresponding to the original data sample at each time point according to a reference positioning system. For example, the reference positioning system can be set up in a specific calibration site to obtain the accurate position information of the carrier equipment in the specific calibration site. For example, the reference positioning system can also be combined with high-precision map data to obtain the accurate position information of the carrier equipment. Thus, label the true position information of the carrier equipment when each data sample is collected. Exemplarily, the true position information of the carrier equipment can be labeled using longitude and latitude coordinates.

[0083] Exemplarily, the method for normalizing the eigenvectors in the three-dimensional data tensor can adopt the min-max normalization method to map the data to the interval [0, 1] for facilitating the training of the deep learning model.

[0084] The composition method of the deep learning model can include various ways. Exemplarily, the deep learning model can include a hybrid architecture composed of a Convolutional Neural Networks (abbreviated as CNN) and a Long Short-Term Memory (abbreviated as LSTM).

[0085] Exemplarily, the input format of the input layer of the hybrid architecture is: a three-dimensional data tensor with a shape of (time series length, eigenvector dimension), and the input layer is used to receive the preprocessed three-dimensional data tensor.

[0086] Exemplarily, the convolutional layer of the hybrid architecture is multiple convolutional kernels with a size of (3, eigenvector dimension), and the convolution step of the convolutional layer is set to a set step. For example, the number of convolutional layers is 32. Exemplarily, the above set step can be 1. Thus, the local features of the sample data are extracted through the convolutional layer.

[0087] Exemplarily, the padding method of the hybrid architecture is the SAME padding method, that is, the "same" padding method is adopted. Of course, in addition to this, other padding methods can also be adopted.

[0088] Exemplarily, the activation function of the hybrid architecture is a linear activation function (ReLU function). Of course, other activation functions can also be adopted.

[0089] Exemplarily, the maximum layer pooling window size of the hybrid architecture is (2, 1), and the long short-term memory network sets multiple hidden units. Exemplarily, the number of neurons in the fully connected layer of the hybrid architecture is determined based on the positioning accuracy requirement. The max pooling layer is used to reduce the data dimension, and the pooling window size is (2, 1). Then the data output by the pooling layer is flattened and input into the LSTM layer. The LSTM layer is used to learn the time series features of the data, and a certain number of hidden units (such as 64) are set. Finally, the fully connected layer is connected. The number of neurons in the fully connected layer is determined according to the positioning accuracy requirement (such as 2 neurons for outputting longitude and latitude coordinates), and a linear activation function is used.

[0090] Exemplarily, refer to Figure 3, a loss function and an optimizer can also be defined. Specifically, the mean squared error function (MSE) is used as the loss function to measure the difference between the first predicted position output by the positioning model and the true position. Exemplarily, the Adam optimizer can be selected as the optimizer, which can adaptively adjust the learning rate to accelerate the training convergence speed of the positioning model and improve stability.

[0091] There are various ways to train a deep learning model based on data samples to obtain a positioning model. Exemplarily, the reliable positioning method for the carrying equipment may further include the following steps:

[0092] Divide the data samples into a training set, a validation set, and a test set;

[0093] Use the training set to perform multiple rounds of training on the deep learning model, and after each round of training, use the validation set to evaluate the loss value of the deep learning model;

[0094] Stop training when the loss value no longer decreases or starts to increase for a continuously set number of rounds, and use the deep learning model obtained after the training ends as the positioning model;

[0095] Use the test set to evaluate the prediction error metrics of the positioning model.

[0096] For example, when dividing the data, the data samples can be divided into a training set, a validation set, and a test set according to a set ratio. For example, the labeled data samples can be divided into a training set, a validation set, and a test set according to the ratio of 70%, 20%, and 10%. Of course, it should be noted that the set ratio is not limited to the ratio of 70%, 20%, and 10% shown above. In addition, other preset ratios can also be used.

[0097] During the model training process, use the training set to train the constructed deep learning model. Input the training data into the deep learning model, calculate the first predicted position through forward propagation, then calculate the loss value according to the MSE loss function, and then update the parameters of the deep learning model using the backpropagation algorithm. During the training process, set a certain number of training rounds (such as 100 rounds), and evaluate the performance of the deep learning model on the validation set after each round of training, and observe the change of the loss value of the validation set. When the loss value of the validation set no longer decreases or starts to increase for several consecutive rounds (such as 5 rounds), stop training in advance to prevent overfitting.

[0098] During the model evaluation process, use the test set to evaluate the performance of the trained positioning model, calculate the prediction error metrics of the positioning model, such as the mean absolute error (MAE), root mean square error (RMSE), etc., to evaluate the positioning accuracy and reliability of the positioning model.

[0099] As described above, refer to Figure 3During the positioning process of the transportation equipment, satellite positioning data, Bluetooth positioning data, and environmental data can be collected in real time through a satellite locator, a Bluetooth locator, and environmental sensors. After cleaning, formatting, and normalizing the collected satellite positioning data, Bluetooth positioning data, and environmental data according to the data preprocessing steps in the above positioning model training process, a feature vector in a format suitable for the positioning model is obtained. The feature vector is input into the trained positioning model, and the positioning model outputs the predicted position coordinates as the first predicted position of the transportation equipment.

[0100] Next, referring to Figure 1 , a fusion positioning strategy is also executed to calculate the second predicted position of the transportation equipment based on the satellite positioning data and the Bluetooth positioning data. For example, a fusion algorithm such as, but not limited to, weighted average or Kalman filter can be used to calculate the second predicted position of the transportation equipment based on the satellite positioning data and the Bluetooth positioning data.

[0101] Next, referring to Figure 1 , the first predicted position predicted by the deep learning model and the second predicted position calculated by the traditional fusion algorithm are also fused. Specifically, based on the prediction error index of the positioning model, the first weight coefficient corresponding to the first predicted position and the second weight coefficient corresponding to the second predicted position are determined.

[0102] Exemplarily, if the mean absolute error and the root mean square error in the prediction error index are small, it indicates that the accuracy and reliability of the first predicted position are relatively high, and the first weight coefficient can be set larger. Correspondingly, the second weight coefficient can be set smaller. If the mean absolute error and the root mean square error in the prediction error index are large, it indicates that the accuracy and reliability of the first predicted position are relatively low, and the first weight coefficient can be set smaller. Correspondingly, the second weight coefficient can be set larger. This is to improve the accuracy and reliability of the final predicted position.

[0103] Next, referring to Figure 1 , based on the first predicted position, the first weight coefficient, the second predicted position, and the second weight coefficient, the final predicted position of the transportation equipment is determined.

[0104] Regarding the method of determining the final predicted position of the transportation equipment based on the first predicted position and the second predicted position, various methods can be adopted.

[0105] Exemplarily, the sum of the first weight coefficient and the second weight coefficient can be equal to 1. At this time, the method of determining the final predicted position of the transportation equipment based on the first predicted position, the first weight coefficient, the second predicted position, and the second weight coefficient can include the following steps:

[0106] The sum of the product of the first longitude in the first predicted position and the first weight coefficient and the product of the second longitude in the second predicted position and the second weight coefficient is used as the final predicted longitude in the final predicted position;

[0107] The sum of the product of the first latitude in the first predicted position and the first weight coefficient and the product of the second latitude in the second predicted position and the second weight coefficient is used as the final predicted latitude in the final predicted position.

[0108] By adopting the above method, the second predicted position calculated by the traditional fusion algorithm can be corrected using the first predicted position predicted by the deep learning model, so as to obtain a final predicted position closer to the real position, improving the positioning accuracy and reliability.

[0109] Exemplarily, according to the performance evaluation results of the positioning model in the test phase, the first weight coefficient w corresponding to the first predicted position output by the positioning model can be determined model , and the second weight coefficient w corresponding to the second predicted position calculated by the traditional fusion algorithm trad =1 - w model .

[0110] The calculation formulas for the longitude coordinate (x final ) and latitude coordinate (y final ) of the final predicted position are:

[0111] x final = w model* x model + w trad* x trad

[0112] y final = w model * y model + w trad * y trad

[0113] where (x model， y model ) are the longitude and latitude coordinates of the first predicted position output by the positioning model, and (x trad， y trad ) are the longitude and latitude coordinates of the second predicted position calculated by the traditional fusion algorithm.

[0114] Exemplarily, referring to Figure 3 , this positioning method further includes:

[0115] When the environmental sensor detects that the carrier equipment is in a stationary state and the change range of the first predicted position obtained based on the positioning model within a set time period does not exceed the set range, the sampling frequency of the satellite locator and the Bluetooth locator is reduced from the first sampling frequency to the second sampling frequency, or the satellite locator and the Bluetooth locator are controlled to switch to a sleep mode, so that the satellite locator and the Bluetooth locator execute a low power consumption strategy, thereby reducing energy consumption;

[0116] When the environmental sensor detects that the carrying equipment starts to move again, the sampling frequency of the satellite locator and the Bluetooth locator is restored to the first sampling frequency, so as to use the positioning model to quickly correct the initial position of the awakened carrying equipment to ensure the continuity and accuracy of positioning.

[0117] Exemplarily, the working modes of the satellite locator and the Bluetooth locator can be dynamically adjusted according to the motion state and positioning environment of the carrying equipment. When the carrying equipment is in a stationary state and the first predicted position output by the positioning model has not changed significantly (the amplitude of the change does not exceed the set range) within a set time (such as 10 minutes), the sampling frequency of the satellite locator and the Bluetooth locator is reduced or switched to a low-power sleep mode. The motion state of the carrying equipment is detected by environmental sensors (such as accelerometers and gyroscopes). When it is detected that the carrying equipment starts to move again, the satellite locator and the Bluetooth locator are awakened and restored to the normal first sampling frequency.

[0118] During the daily transportation of the carrying equipment, the fusion positioning and low-power consumption strategy are run in real time. The precise location information of the carrying equipment is obtained based on the first predicted position predicted by the positioning model and the second predicted position calculated by the traditional algorithm. The power consumption of the positioning module is dynamically adjusted according to the status of the carrying equipment to achieve efficient, low-power and high-precision positioning of hazardous chemical tank trucks.

[0119] Example 2

[0120] In addition, another embodiment of the present invention further provides a trusted positioning device for carrying equipment, which is provided with at least a satellite locator, a Bluetooth locator and an environmental sensor. The positioning device includes: a storage medium and a processor, the storage medium stores a computer program run by the processor, and when the computer program is run by the processor, the processor executes any one of the above-mentioned trusted positioning methods for carrying equipment.

[0121] Among them, the processor may include processor devices with logical operation functions such as, but not limited to, single-chip microcomputers, MCUs (microcontroller units), etc. The storage medium may, for example, include a memory card of a smart phone, a storage component of a tablet computer, a hard disk of a personal computer, a read-only memory (ROM), an erasable programmable read-only memory (EPROM), a portable compact disc read-only memory (CD-ROM), a USB memory, or any combination of the above storage media. The computer-readable storage medium may be any combination of one or more computer-readable storage media.

[0122] This application has been illustrated by the above embodiments. However, it should be understood that the above embodiments are only for the purpose of exemplification and illustration, and are not intended to limit this application to the scope of the described embodiments. In addition, those skilled in the art can understand that this application is not limited to the above embodiments, and more variations and modifications can be made according to the teachings of this application, and these variations and modifications all fall within the scope claimed by this application. The protection scope of this application is defined by the appended claims and their equivalent scope.

Claims

1. A reliable positioning method for transport equipment, characterized in that: The carrier equipment is provided with at least a satellite locator, a Bluetooth locator and an environmental sensor, and the carrier equipment trusted positioning method includes: Acquire satellite positioning data collected by the satellite locator, Bluetooth positioning data collected by the Bluetooth locator, and environmental data collected by the environmental sensor; Input the satellite positioning data, the Bluetooth positioning data and the environmental data into a positioning model to obtain a first predicted position of the carrying equipment; wherein the positioning model is obtained by training a deep learning model based on data samples, and the data samples include: satellite positioning data samples collected based on the satellite locator, Bluetooth positioning data samples collected based on the Bluetooth locator, and environmental data samples collected based on the environmental sensor; Calculate and obtain a second predicted position of the carrying equipment based on the satellite positioning data and the Bluetooth positioning data; Determining a first weight coefficient corresponding to the first predicted position and a second weight coefficient corresponding to the second predicted position based on a prediction error index of the positioning model; Based on the first predicted position, the first weight coefficient, the second predicted position and the second weight coefficient, the final predicted position of the carrying equipment is determined.

2. The method for reliable positioning of a carrier equipment according to claim 1, characterized in that: The sum of the first weight coefficient and the second weight coefficient is equal to 1; The determining the final predicted position of the carrying equipment based on the first predicted position, the first weight coefficient, the second predicted position and the second weight coefficient comprises: taking the sum of the product of the first longitude in the first predicted position and the first weight coefficient and the product of the second longitude in the second predicted position and the second weight coefficient as the final predicted longitude in the final predicted position; The product of the first latitude in the first predicted position and the first weight coefficient, and the sum of the product of the second latitude in the second predicted position and the second weight coefficient are used as the final predicted latitude in the final predicted position.

3. The reliable positioning method for transport equipment according to claim 1, characterized in that: Also includes: When the environmental sensor detects that the carrying equipment is in a stationary state and the change amplitude of the first predicted position obtained based on the positioning model within a set time period does not exceed a set range, the sampling frequency of the satellite locator and the Bluetooth locator is reduced from the first sampling frequency to the second sampling frequency, or the satellite locator and the Bluetooth locator are controlled to switch to a sleep mode; When the environmental sensor detects that the carrying equipment starts to move again, the sampling frequencies of the satellite locator and the Bluetooth locator are restored to the first sampling frequency.

4. The reliable positioning method for carrying equipment according to claim 1, characterized in that: The satellite positioning data and the satellite positioning data samples both include: at least one of satellite signal strength, satellite elevation angle and number of satellites; and / or, The Bluetooth positioning data and the Bluetooth positioning data sample both include: at least one of Bluetooth signal strength, Bluetooth signal arrival angle, and connection status between the Bluetooth locator and surrounding Bluetooth beacons; and / or, The environmental sensor includes at least one of an accelerometer, a gyroscope, a barometer and a camera device, and the environmental data and the environmental data samples include: acceleration of the carrying equipment, angular velocity of the carrying equipment, air pressure changes detected by a barometer, and at least one of indoor and outdoor signs of the environment surrounding the carrying equipment.

5. The reliable positioning method for carrying equipment according to claim 1, characterized in that: Also includes: The satellite positioning data samples, the Bluetooth positioning data samples and the environmental data samples when the carrying equipment is running in different transportation environments are collected at a set sampling frequency to form raw data samples arranged in time series; Cleaning the original data sample; Calculate the feature vector of the original data sample at each time point after cleaning, and combine the feature vectors of multiple time points into a three-dimensional data tensor of shape (time series length, feature vector dimension); wherein different types of data in the original data sample are arranged in sequence in the feature vector; According to the reference positioning system, marking the real position information corresponding to the original data sample at each time point; The feature vector in the three-dimensional data tensor is normalized, and the three-dimensional data tensor after the normalization is used as the data sample.

6. The method for reliable positioning of a carrier equipment according to claim 1, characterized in that: Also includes: Dividing the data sample into a training set, a validation set and a test set; Perform multiple rounds of training on the deep learning model using the training set, and after each round of training, use the validation set to evaluate the loss value of the deep learning model; When the loss value stops decreasing or starts to increase after a set number of consecutive rounds, the training is stopped, and the deep learning model obtained after the training is completed is used as the positioning model; The prediction error indicator of the positioning model is evaluated using the test set.

7. The reliable positioning method for transport equipment according to claim 1, characterized in that: The deep learning model includes: a hybrid architecture based on a convolutional neural network and a long short-term memory network; The input format of the input layer of the hybrid architecture is: a three-dimensional data tensor with a shape of (time series length, feature vector dimension); and / or, The convolution layer of the hybrid architecture is a plurality of convolution kernels of size (3, feature vector dimension), and the convolution step size of the convolution layer is set to a set step size; and / or, The filling mode of the hybrid architecture is the SAME filling mode; and / or, The activation function of the hybrid architecture is a linear activation function; and / or, The maximum layer pooling window size of the hybrid architecture is (2, 1), and the long short-term memory network is provided with a plurality of hidden units; and / or, The number of neurons in the fully connected layer of the hybrid architecture is determined based on the positioning accuracy requirement; and / or, The hybrid architecture loss function is defined as a mean square error function, and the optimizer of the hybrid architecture is defined as an Adam optimizer.

8. A reliable positioning device for transport equipment, characterized in that: The carrier equipment is provided with at least a satellite locator, a Bluetooth locator and an environmental sensor, and the carrier equipment trusted positioning device includes: A storage medium and a processor, wherein the storage medium stores a computer program executed by the processor, and when the computer program is executed by the processor, the processor executes the trusted positioning method for carrier equipment as described in any one of claims 1 to 7.

Images