Industrial Internet of Things monitoring system and method based on cloud edge collaboration

Abstract

The invention discloses an industrial Internet of Things monitoring system based on cloud edge collaboration. The industrial Internet of Things monitoring system comprises an equipment end and a system cloud end, the equipment end comprises a plurality of sensors deployed on processing equipment and a communication module, and the communication module is used for uploading monitoring data collected by the plurality of sensors to the system cloud end; the system cloud comprises a data communication layer, a data storage layer, a data analysis layer and a data visualization layer. The invention also provides an industrial Internet of Things monitoring method. The system provided by the invention can realize the dynamic optimization deployment of the distributed workflow.

Description

Technical Field

[0001] The present invention belongs to the field of intelligent manufacturing, and in particular relates to an industrial Internet of Things monitoring system and method based on cloud-edge collaboration. Background Art

[0002] Large-scale equipment monitoring technology within the context of the Industrial Internet of Things (IIoT) serves as a core pillar of modern intelligent manufacturing systems. Its research value lies not only in enabling transparent control of production processes but also in driving the digital transformation of cross-industry manufacturing paradigms. By building a real-time monitoring network covering heterogeneous equipment clusters, the system enables multi-dimensional perception and predictive analysis of operational status, providing a dynamic basis for decision-making in process optimization. However, practical deployment of this technology still faces multiple technical bottlenecks. First, the widespread heterogeneity of equipment in industrial sites manifests itself in differences in sensor interface protocols, varying data acquisition frequencies, and non-uniform signal dimensions, placing strict compatibility requirements on the construction of a unified analytical framework. Second, nonlinear coupling mechanisms between process parameters can lead to distortions in feature space. For example, the complex interaction between thermodynamic parameters and mechanical stresses in material processing directly reduces the prediction confidence of traditional linear models. Third, the dynamic characteristics of energy-intensive industrial systems, such as continuous-flow chemical plants, exhibit significant time dependence. Therefore, a time-series graph neural network, constructed by integrating historical equipment operating data, is required to effectively model long-range correlations between process parameters and compensate for inherent sensor transmission delays.

[0003] Patent document CN113822496A discloses a method for online optimization of heating modes and parameters of a multi-unit thermal power plant, including: step S1, setting a multi-heating mode combination scheme when the heating unit participates in deep peak-shaving coordinated operation; step S2, using mechanism modeling and data identification methods to construct a digital twin model of the unit's multi-heating mode coordinated operation; step S3, constructing a unit multi-mode coordinated operation evaluation model; step S4, predicting the heating load and power generation load of the thermal power plant; step S5, calculating the economic indicators and peak-shaving capacity under the power plant-level multi-heating mode combination; step S6, performing optimization analysis and optimization calculation of the multi-heating mode for the peak-shaving capacity, economic indicators and evaluation indicators under different operating schemes of the multi-mode peak-shaving of the heating unit, selecting the multi-mode coordinated operation mode and the best combination scheme of the multi-mode peak-shaving coordinated operation, and online guiding the unit to optimize power generation and heating operation.

[0004] Patent document CN119887125A discloses a data analysis and collaboration system and method for an industrial Internet of Things main service platform, including: a user platform, a service platform, a management platform, a sensor network platform, and an object platform; the object platform includes multiple production devices, each of which is equipped with a sensing unit and a control unit, wherein the sensing unit is used to collect physical quantities of equipment operation and product processing parameters, perform local preliminary screening and caching, mark urgent key data, remove invalid data, and transmit it to the sensor network platform to form preliminary perception data; the sensor network platform includes multiple sensor network sub-platforms and corresponding sensor network databases. Summary of the Invention

[0005] The purpose of the present invention is to provide an industrial Internet of Things monitoring system and method based on cloud-edge collaboration, which realizes dynamic optimization deployment of distributed workflows through containerization encapsulation and microservice orchestration technology.

[0006] In order to achieve the first object of the present invention, the following technical solutions are provided: an industrial Internet of Things monitoring system based on cloud-edge collaboration, which includes a device side and a system cloud;

[0007] The device side includes a variety of sensors deployed on the processing equipment and a communication module, and the communication module is used to upload the monitoring data collected by the various sensors to the system cloud;

[0008] The system cloud includes a data communication layer, a data storage layer, a data analysis layer, and a data visualization layer;

[0009] The data communication layer is used to receive the monitoring data uploaded by the communication module and send the received monitoring data to the data storage layer for storage;

[0010] The data storage layer includes a real-time database and a backup database, wherein the real-time database processes the received monitoring data in a time series format to obtain corresponding real-time monitoring data and stores the real-time monitoring data;

[0011] The data analysis layer performs data filling operations on the real-time monitoring data of sensors of different dimensions in the real-time database, and inputs the results after the data filling operation into the pre-trained prediction model for prediction, so as to obtain the prediction results of the corresponding time and store them in the backup database;

[0012] The data visualization layer retrieves the real-time monitoring data and the corresponding prediction results from the data storage layer according to the selected processing equipment, and performs visualization output.

[0013] Specifically, the device uses EtherCAT and / or RS485 communication protocols to collect monitoring data from multiple sensors.

[0014] Specifically, the communication module uses the MQTT communication protocol to upload the monitoring data to the system cloud.

[0015] Specifically, the data communication layer is constructed based on the EMQX service framework.

[0016] Specifically, the data storage layer is constructed based on a time series database architecture, and stores the input monitoring data and corresponding prediction results in a time series format.

[0017] Specifically, the prediction model includes a feature extraction module and a prediction module. The feature extraction module is used to extract data features of the monitoring data after data filling is completed. The prediction module makes predictions based on the input data features to obtain the operating status of the processing equipment at the next moment.

[0018] Specifically, the prediction model is constructed using a long short-term memory neural network with an encoding-decoding architecture.

[0019] Specifically, the data filling operation includes one or more of zero filling, average value filling, mirror filling or circular filling.

[0020] In order to achieve the second purpose of the present invention, the following technical solution is provided: an industrial Internet of Things monitoring method is implemented by the industrial Internet of Things monitoring system based on cloud-edge collaboration provided above.

[0021] Compared with the prior art, the present invention has the following beneficial effects:

[0022] The present invention is based on a cloud-edge collaborative computing platform built on the Kubernetes engine. It integrates and analyzes various types of sensor data through data filling combined with prediction models, thereby realizing the large-scale deployment of the Industrial Internet of Things. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] Figure 1 Schematic diagram of the industrial Internet of Things monitoring system based on cloud-edge collaboration provided in this embodiment;

[0024] Figure 2 A schematic diagram of an industrial Internet of Things monitoring method provided in this embodiment. DETAILED DESCRIPTION

[0025] In order to make the purpose, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. The components of the embodiments of the present invention generally described and shown in the drawings herein can be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of the present invention provided in the drawings is not intended to limit the scope of the invention claimed for protection, but merely represents selected embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention.

[0026] like Figure 1 As shown in the figure, this example provides an industrial Internet of Things monitoring system based on cloud-edge collaboration. Taking the injection molding barrel heating process as an example, in large-scale monitoring, due to the different number of barrel temperature sections and the number of sensors of different injection molding equipment, the dimensions and structures of the collected data are different. The following steps are used to collect and analyze data.

[0027] The equipment collects real-time heating power and temperature sensor signals. For example, for an injection molding barrel with five temperature ranges, the real-time monitoring data includes five heating power and five temperature readings. This real-time monitoring data is transmitted to a PLC controller via industrial fieldbus protocols such as EtherCAT. A Raspberry Pi edge computing node then retrieves historical temperature curves and power operating data from the controller's data registers via Modbus / TCP or EtherCAT. After local preprocessing, the collected data is transmitted to the local area network via the MQTT IoT protocol.

[0028] The data communication layer is implemented through the EMQX service. A publish / subscribe model is used to establish a communication link between edge devices (publishers) and cloud monitoring systems (subscribers). Data is transmitted via designated topics and then pushed to the storage layer. Through the EMQX service, the data communication layer can capture data published on the local area network and transfer it to the cloud monitoring system.

[0029] The data storage layer uses a time-series database architecture and selects InfluxDB, which is optimized for high-throughput read and write. When receiving real-time monitoring data, it first adds a timestamp to the data and stores the data in chronological order before and after the timestamp. All data is stored in the Influx DB database.

[0030] The data analysis layer deploys data population algorithms and prediction models. For example, if the monitored production equipment is an injection molding machine with 4, 5, or 6 temperature ranges, and the real-time monitoring data collected contains 8, 10, or 12 feature data (including heating power and temperature data), all data must first be populationd to 12 features. Then, using the heating power and temperature data from several historical steps, the temperature data for several future steps must be predicted.

[0031] First, extract the real-time monitoring data from the database, and use data filling methods such as zero filling, average filling, mirror filling or loop filling to fill the data with different dimensions and structures to the same dimension. Let the original data be X, X is a three-dimensional matrix data, and its three dimensions are: batch size n batch , historical time length T his , the number of data features l val .

[0032] Since the number of features of monitoring data from different production equipment is different, that is, val Different, the last dimension of X needs to be filled to a uniform value, i.e. l max .

[0033] The zero-filling method is to assign all the feature elements that need to be filled to 0. The data filling process of zero filling can be expressed by formula (1):

[0034]

[0035] Among them, X pad,zero is the data obtained after zero filling, X pad,zero [:,:,i] means that the first two dimensions are arbitrary values, that is, any batch and historical time step, and the third dimension is i, that is, the i-th feature.

[0036] The average value filling method is to assign all feature elements that need to be filled to the average value of all features in the batch and the historical time step, as shown in formula (2) and formula (3):

[0037]

[0038] Among them, μ[b,t] is the average value of all features at the b-th batch and the t-th historical time step, X pad,average The data is obtained after filling with the mean value.

[0039] The mirror filling method is to assign the value of the original data to the feature value to be filled in the mirror order, as shown in formula (4):

[0040]

[0041] Among them, X pad,reflectThe data obtained after mirror filling is in X[:,:,l val -i], when l val - When i is less than zero, it indicates the last few elements of X.

[0042] The cyclic filling method is to assign the values of the original data to the feature values to be filled in the original order, as shown in formula (5):

[0043]

[0044] Among them, X pad,circualr The data obtained after loop filling.

[0045] After completing the data filling operation, the filled data X pad Input into the prediction model. This embodiment uses a long short-term memory neural network (LSTM) with an encoder-decoder architecture. The encoder-decoder architecture includes an encoder and a decoder. The encoder encodes the input data into a fixed-dimensional context vector, and the decoder generates the final prediction value based on the context vector. In this embodiment, both the encoder and the decoder are LSTM neural networks. In the encoder, the LSTM neural network performs calculations for each time step, as shown in equations (5) to (7).

[0046] i t =σ(W i [h t-1 ,X t ]+b i ) (5)

[0047] f t =σ(W f [h t-1 ,X t ]+b f ) (6)

[0048] o t =σ(W o [h t-1 ,X t ]+b o ) (7)

[0049] Among them, i t 、f t 、o t They are input gate, forget gate, and output gate respectively, t is the historical time step, W i 、W f 、W o is the trainable network weight, b i 、b f 、b o is the bias vector, h t-1 is the hidden layer state, Xt is the input data of the current historical time step, σ represents the sigmoid activation function, and its expression is

[0050] The neuron state and hidden layer state update methods of the LSTM neural network are shown in Equation (8) and Equation (9):

[0051] c t =f t ⊙c t-1 +i t ⊙tanh(W c [h t-1 ,X t ]+b c ) (8)

[0052] h t =o t ⊙tanh(c t ) (9)

[0053] Among them, c t is the neuron state, h t is the hidden layer state, W c and b c are the learnable network weights and bias vectors, respectively, and ⊙ represents the multiplication of corresponding elements. Dropout is used at the end of the encoder to prevent overfitting of the network training. The final neuron state and the hidden layer state Passed to the decoder as a context vector.

[0054] In the decoder, the neuron state and hidden layer state of the LSTM network are first initialized according to the context vector, as shown in Equations (10) and (11):

[0055]

[0056] in, are the initial neuron state and hidden layer state of the LSTM network in the decoder, respectively.

[0057] In the decoder, the final hidden layer state h of the decoder at the previous time step is dec [-1] is concatenated with the input of the decoding layer, as shown in formula (12):

[0058] X dec,cat =Concatenate(h dec [-1],X dec ,-1) (12)

[0059] Among them, X dec,cat is the decoder input after splicing, Xdec For the original decoder input, the Concatenate(a,b,-1) function represents the concatenation operation of a and b along the last dimension.

[0060] The calculation process of the LSTM network is shown in Equations (5) to (9). Finally, the final prediction value is obtained through the full connection layer and the softmax activation function, as shown in Equation (13):

[0061]

[0062] in, is the final predicted value, y dec is the output value of the decoder, W y and b y are the learnable network weights and bias vectors of the fully connected layer, respectively. The softmax function is calculated as

[0063] The loss function is calculated based on the predicted value. The loss function forms include mean absolute error, mean square error, etc. At the same time, the masked loss function is used. When calculating the prediction accuracy, only the data of the original length is considered for comparison. The calculation method is shown in Equation (14) and Equation (15):

[0064]

[0065]

[0066] Among them L mask,MAE 、L mask,MSE They represent the mean absolute error of the mask and the mean squared error of the mask loss functions, respectively.

[0067] During the training process, the learnable parameters are continuously updated. If the number of training times reaches the set value n, epoch , or the number of times the loss function value does not decrease during continuous training reaches the set value n patience , stop training. The final model can achieve accurate prediction of time series.

[0068] After obtaining the predicted data, the prediction model stores it in the Influx DB database in chronological order and adds a timestamp label to the database.

[0069] This embodiment also provides an industrial Internet of Things monitoring method, such as Figure 2 As shown, the industrial Internet of Things monitoring system based on cloud-edge collaboration provided by the above embodiment is implemented.

[0070] In addition, the terms "upper", "lower", "inner", "outer", "front", and "back" are used for descriptive purposes only and should not be understood as indicating or implying relative importance. Unless otherwise specifically stated, the relative steps, numerical expressions, and values of the components and steps described in these embodiments do not limit the scope of the present invention.

[0071] Of course, the above description is only a specific embodiment of the present invention and is not intended to limit the scope of implementation of the present invention. Any equivalent changes or modifications made based on the structure, features and principles described in the scope of the patent application of the present invention should be included in the scope of the patent application of the present invention.

[0072] Finally, it should be noted that the above-described embodiments are only specific implementation methods of the present invention, which are used to illustrate the technical solutions of the present invention, rather than to limit them. The scope of protection of the present invention is not limited thereto. Although the present invention has been described in detail with reference to the above-described embodiments, those skilled in the art should understand that any person skilled in the art can modify or easily conceive of changes to the technical solutions described in the above-described embodiments within the technical scope disclosed by the present invention, or replace some of the technical features therein with equivalents. Such modifications, changes, or replacements do not deviate from the spirit and scope of the technical solutions of the embodiments of the present invention, and should be included in the scope of protection of the present invention. Therefore, the scope of protection of the present invention shall be subject to the scope of protection of the claims.

Claims

1. An industrial Internet of Things monitoring system based on cloud-edge collaboration, characterized by: Including device side and system cloud; The device side includes a variety of sensors deployed on the processing equipment and a communication module, and the communication module is used to upload the monitoring data collected by the various sensors to the system cloud; The system cloud includes a data communication layer, a data storage layer, a data analysis layer, and a data visualization layer; The data communication layer is used to receive the monitoring data uploaded by the communication module and send the received monitoring data to the data storage layer for storage; The data storage layer includes a real-time database and a backup database, wherein the real-time database processes the received monitoring data in a time series format to obtain corresponding real-time monitoring data and stores the real-time monitoring data; The data analysis layer performs data filling operations on the real-time monitoring data of sensors of different dimensions in the real-time database, and inputs the results after the data filling operation into the pre-trained prediction model for prediction, so as to obtain the prediction results of the corresponding time and store them in the backup database; The data visualization layer retrieves the real-time monitoring data and the corresponding prediction results from the data storage layer according to the selected processing equipment, and performs visualization output.

2. The industrial Internet of Things monitoring system based on cloud-edge collaboration according to claim 1 is characterized in that: The device uses EtherCAT and / or RS485 communication protocols to collect monitoring data from multiple sensors.

3. The industrial Internet of Things monitoring system based on cloud-edge collaboration according to claim 1 is characterized in that: The communication module uses the MQTT communication protocol to upload monitoring data to the system cloud.

4. The industrial Internet of Things monitoring system based on cloud-edge collaboration according to claim 1 is characterized in that: The data communication layer is built based on the EMQX service framework.

5. The industrial Internet of Things monitoring system based on cloud-edge collaboration according to claim 1 is characterized in that: The data storage layer is constructed based on a time series database architecture, and stores the input monitoring data and corresponding prediction results in a time series format.

6. The industrial Internet of Things monitoring system based on cloud-edge collaboration according to claim 1 is characterized in that: The prediction model includes a feature extraction module and a prediction module. The feature extraction module is used to extract data features of the monitoring data after data filling is completed. The prediction module makes predictions based on the input data features to obtain the operating status of the processing equipment at the next moment.

7. The industrial Internet of Things monitoring system based on cloud-edge collaboration according to claim 1 or 6, characterized in that: The prediction model is constructed using a long short-term memory neural network with an encoding-decoding architecture.

8. The industrial Internet of Things monitoring system based on cloud-edge collaboration according to claim 1 is characterized in that: The data filling operation includes one or more of zero filling, average value filling, mirror filling or circular filling.

9. An industrial Internet of Things monitoring method, characterized in that: This is achieved through the industrial Internet of Things monitoring system based on cloud-edge collaboration as described in any one of claims 1 to 8.

Images