Algorithm model creation system based on configuration technology

Abstract

The invention provides an algorithm model creation system based on a configuration technology, which comprises the following steps of: 1, setting configuration model nodes which comprise variable class nodes, function class nodes and end nodes; 2, connecting nodes to complete configuration model logic; 3, running the configuration model logic to obtain test data, and uploading the test data to complete the test of the configuration model logic to obtain a test result; 4, judging whether the test result can accurately hit the fault or give an early warning or not, and if the test result accurately hit the fault or give an early warning, entering S5; if the fault is not accurately hit or early warning is not performed, returning to the step 1, and repeating the steps 1-4 until entering the step 5; 5, configuration model logic is built, an algorithm model building system based on the configuration technology is completed, and the system comprises the configuration model logic, background service logic and model dispatch.According to the method, health management is conducted on the equipment, the health condition of the equipment is managed and predicted, and conversion of the equipment from planned repair to state repair is achieved.

Description

Technical Field

[0001] The invention belongs to the technical field of rail transit system operation and maintenance, and in particular relates to an algorithm model creation system based on configuration technology. Background Art

[0002] Safety and reliability are the most basic and important requirements for rail operation and maintenance. With the development of technology, rail transit operation and maintenance are becoming increasingly complex, and the diversity, integration and intelligence of data are constantly improving. In modern train operation and maintenance, configuration creation model logic has become a trend in train operation and maintenance, and dynamic creation of fault or early warning monitoring logic has gradually become an important means of train operation and maintenance.

[0003] Algorithm model creation is currently rarely used in actual train operation and maintenance. Most of the methods used are to first determine a fixed fault or warning logic, and the service provider writes the monitoring service in advance to implement specific fault or warning monitoring during train operation. However, as the technology used on trains becomes more and more advanced, some of the original fixed monitoring logics can no longer meet the requirements of operation and maintenance. Summary of the invention

[0004] The object of the present invention is to provide an algorithm model creation system based on configuration technology, characterized in that it includes the following steps:

[0005] S1: Set the configuration model nodes, which include variable nodes, function nodes, user-defined function nodes and end nodes;

[0006] S2: Connect each node to complete the configuration model logic;

[0007] S3: upload test data, run the configuration model logic, complete the configuration model logic test through the test data, and obtain the test results;

[0008] S4: Determine whether the test result can accurately hit the fault or warning. If it can accurately hit the fault or warning, enter S5; if it cannot accurately hit the fault or warning, return to S1 and repeat S1-S4 until entering S5;

[0009] S5: Build the configuration model logic and complete the algorithm model creation system based on configuration technology.

[0010] Furthermore, in S1, the variable class nodes include variable nodes and constant nodes, wherein the variable is the data of each device when the train is running, and the constant is a specific value set.

[0011] Further, in S1, the function class node includes a data processing method and a data acquisition module. The data processing method includes arithmetic operations, comparison operations, logical operations, mathematical functions, cutting operations and aggregation functions. The data acquisition module includes a counter node and a timer node. The counter node specifically sets the start condition and the end condition, and collects a set of data according to the set conditions. The timer node specifically sets the start condition and the collection duration or data volume, and then collects a set of data according to the setting.

[0012] Custom function nodes are specifically for complex data processing or calculation needs. When function nodes cannot meet user needs, users can set parameters and write Python code to connect the corresponding data to the parameter items of the code, thereby realizing data processing, collection or verification.

[0013] Furthermore, in S1, the end node specifically sets the alarm level and alarm type of the abnormal situation. When the data meets the set abnormal situation, the system reports a fault or warning according to the information set by the end node.

[0014] Furthermore, the end node responds to different configuration model types by setting different data.

[0015] Furthermore, in S2, the connection of nodes is as follows: in the modeled membrane editing page, by dragging in variable nodes, function nodes or custom function nodes and end nodes in sequence; specifically:

[0016] Data mapping is achieved by dragging in variable nodes. After dragging in the variable node and selecting the corresponding data field, the variable node represents the selected device data itself; then drag in the function node or the custom function node to perform corresponding calculations or collections on the data. When dragging in a function node, select the corresponding calculation logic in the function node. When connecting the variable node to the function node, data circulation can be achieved; when dragging in a custom function node, when connecting the variable node, make the variable node correspond to the parameters set by the custom function one by one; finally, drag in the end node and connect the final calculated result node to the end node. The end node judges the result data. When the abnormal conditions set by the end node are met, the system reports a fault or warning according to the information set by the end node.

[0017] Furthermore, S3 is specifically: running the combined model logic, and generating test data in a targeted manner according to the combined model logic, and verifying whether the model logic can accurately hit faults or warnings by uploading the test data.

[0018] Furthermore, in S5, the algorithm model creation system based on configuration technology includes configuration model logic, background service logic and model scheduling.

[0019] Furthermore, the background service logic is as follows: according to the edited configuration model logic, take the end node as the starting point, reason forward, record the execution order of all nodes, obtain the corresponding data field according to the variable node, push the data to the node execution order to start calculating the logic of each node, and when the data flows to the end node, determine whether the current result meets the threshold set by the end node. If it meets the threshold, add a new data result.

[0020] Furthermore, the model scheduling is: periodic execution of the configuration model, the periodicity includes real-time scheduling, interval scheduling and single scheduling;

[0021] Real-time scheduling is specifically: monitoring data in real time and executing model logic immediately when data is available;

[0022] Interval scheduling is specifically: setting a fixed time interval period, within which the configuration model is executed once, and the data content is all the data within the period;

[0023] Single scheduling is specifically: the configuration model is enabled once every 24 hours, and the data content is the data within the 24 hours.

[0024] Compared with the prior art, the beneficial effects of the present invention are mainly reflected in:

[0025] 1. The present invention has a built-in algorithm that manages the health of the equipment and predicts the health of the equipment, thereby realizing the transformation of the equipment from planned maintenance to status maintenance.

[0026] 2. The present invention provides a low-code platform for model engineers, which allows them to build models by simply dragging and dropping, thereby reducing the difficulty of getting started and reducing customers' development costs. BRIEF DESCRIPTION OF THE DRAWINGS

[0027] Figure 1 It is a schematic diagram of the connection of each node of the configuration model logic in the present invention.

[0028] Figure 2 Schematic diagram of the fault or warning process in the present invention DETAILED DESCRIPTION

[0029] The following is a more detailed description of an algorithm model creation system based on configuration technology of the present invention in conjunction with a schematic diagram, in which a preferred embodiment of the present invention is shown. It should be understood that those skilled in the art can modify the present invention described herein and still achieve the beneficial effects of the present invention. Therefore, the following description should be understood as being widely known to those skilled in the art and not as a limitation of the present invention.

[0030] An algorithm model creation system based on configuration technology includes the following steps:

[0031] Step 1: Set up each node and connect them.

[0032] Nodes include: variable nodes, function nodes, user-defined function nodes and end nodes.

[0033] 1.1: Variable nodes are the data source of the monitoring logic in the configuration model. When processing and collecting some data, you need to connect to the variable node first to connect to the real data. It contains different types of data, such as fault variables, state variables, control variables, continuous variables, time variables, and constants. The total white energy node includes two types of nodes: variables and constants. Variables represent the data of each device when the train is running, and constants represent a specific value set.

[0034] When you drag a variable node into the configuration editing page, you can edit the variable node and select different device data for binding; when you drag a constant node, you can manually enter a specific value, and the configuration model will calculate the input value when executing the logic.

[0035] 1.2: Function nodes include some common data processing methods and data collection modules, such as arithmetic operations, comparison operations, logical operations, common mathematical functions, cutting operations, and common aggregation functions. The data collection module has timers and timer nodes, which are nodes that collect a period of time or a specific number of data items according to the set conditions.

[0036] 1.2.1: Arithmetic operations: You can choose addition, subtraction, multiplication, division, remainder and other calculation logic.

[0037] 1.2.2: Comparison operation: Calculate multiple data sources, and you can choose calculation logic such as equal, not equal, greater than, greater than or equal to, less than, less than or equal to.

[0038] 1.2.3: Logical operation: Perform logical judgment on multiple data, you can choose logical AND, logical OR, logical NOT, inversion, etc.

[0039] 1.2.4: Mathematical function: Perform mathematical formula calculations on data, you can choose integral, sine, tangent, cotangent, power function, exponential function, etc.

[0040] 1.2.5: Cutting operation: supports 1 continuous variable array, performs cutting operation on the data array, and outputs multiple groups of continuous variables.

[0041] 1.2.6: Aggregation function: Perform aggregation operations on a continuous variable array. You can choose sum, maximum, minimum, average, count, and variance.

[0042] 1.2.7: Timer node: Set the start condition and end condition, and the node will collect a set of data according to the set conditions.

[0043] 1.2.8: Timer node: Set the start condition and the duration or amount of data to be collected, and the node will collect a set of data according to the settings.

[0044] 1.3: Custom function nodes, users can set parameters and write Python code to process, collect or verify data. It is used for some complex data processing or calculation needs. Users can write code according to their needs and just need to connect the corresponding data to the parameter items of the code.

[0045] When ordinary nodes cannot meet the needs, you can write Python code to meet the needs; set the data parameters that the code needs to receive, and connect the corresponding data source to the parameter position on the model editing page.

[0046] 1.4: End node, set the alarm level and alarm type of abnormal situations; when the data meets the set abnormal situation, the system will report a fault / warning according to the information set in the end node.

[0047] Alarm levels include: early warning and alarm.

[0048] Alarm types include: fault alarm, health warning and life warning.

[0049] Fault alarm: The model reports a device fault, indicating that it has affected the operation of the train.

[0050] Health warning: The model sets different thresholds to detect the current health status of the device

[0051] Life warning: The model sets the life threshold of the equipment. When the life is lower than the set threshold, an early warning is issued.

[0052] Each configuration model can only drag one end node. Different model types require different data to be set in the end node.

[0053] In health alarms and life warnings, the threshold settings are different based on different devices.

[0054] 1.4: Node connection.

[0055] like Figure 1As shown, in the model canvas editing page, you can implement data mapping by dragging in variable nodes. After dragging in the variable node and selecting the corresponding data field, the variable node represents the selected device data itself. After dragging the required device data into the model canvas, you can drag in the function class node or the custom function class node to calculate or collect the data accordingly. When dragging in the function class node, select the corresponding calculation logic in the function class node. When the variable node is connected to the function class node, the data flow can be realized. When the built-in nodes of the system cannot meet the logical requirements of faults / warnings, drag in the custom function class node. The node allows users to define the logic of data that needs to be processed or judged, and define the value of the custom function class node that flows to the next node after processing the data. When using the custom function class node, you need to set which data the node is to process (configurable through the variable node). When the variable node (data node) is connected to the custom function class node, the two nodes establish a connection relationship, and the device data will flow to the custom function class node. When connecting variable nodes, you need to pay attention to the one-to-one correspondence between the variable nodes and the parameters set by the custom function; after the multi-layer function node calculates the data, drag in the end node and connect the final calculated result node to the end node. The end node judges the result data and records the result when it meets the rules set by the end node.

[0056] Step 2: Run the configuration model logic, obtain test data, upload the test data to complete the configuration model logic test, and obtain the test results.

[0057] On the model canvas editing page, after completing the logical monitoring of faults / warnings by dragging in variable nodes, function nodes / custom function nodes, end nodes, etc., you can generate a batch of test data based on the model logic. By uploading the test data, you can verify whether the model logic can accurately hit the faults / warnings. After the test, each node of the model will record the results generated by the test data. You can view the test results of the current node by checking the corresponding node information.

[0058] Step 3: Based on the above test results, determine whether the test results can accurately hit the fault or warning.

[0059] Step 4: Complete the algorithm model creation system based on configuration technology.

[0060] 4.1: Build configuration model logic.

[0061] Refer to step 1.4 to connect each node and complete the construction of the configuration model logic.

[0062] 4.2: Backend service logic.

[0063] According to the edited model logic, take the end node as the starting point, reason forward, record the execution order of all nodes (for the case where there are multiple data collection nodes in a model, determine whether the trigger conditions of all collection nodes are met at the same time), obtain the corresponding data field according to the variable node, push the data to the node execution order to start calculating the logic of each node, and when the data flows to the end node, determine whether the current result meets the threshold set by the end node. If it meets the threshold, add a new data result.

[0064] 4.3: Model scheduling.

[0065] refer to Figure 2 ,The model execution cycle is classified into real time or single or interval.

[0066] 4.3.1: Real-time scheduling: Monitor data in real time and execute model logic immediately when data is available.

[0067] 4.3.2: Interval scheduling: Set a fixed time interval period. During this period, the model is executed once, and the data content is all the data within the period.

[0068] 4.3.3: Single scheduling: The model is enabled once every 24 hours, and the data content is the data of the past 24 hours.

[0069] The above is only a preferred embodiment of the present invention and does not limit the present invention in any way. Any technician in the relevant technical field, without departing from the scope of the technical solution of the present invention, makes any form of equivalent replacement or modification to the technical solution and technical content disclosed in the present invention, which does not depart from the content of the technical solution of the present invention and still falls within the protection scope of the present invention.

Claims

1. An algorithm model creation system based on configuration technology, characterized in that: The steps include: S1: Setting configuration model nodes, wherein the configuration model nodes include variable type nodes, function type nodes, user-defined function type nodes and end nodes; S2: Connect nodes to complete the configuration model logic; S3: upload test data, run the configuration model logic, complete the configuration model logic test through the test data, and obtain the test results; S4: Determine whether the test result can accurately hit the fault or warning. If it accurately hits the fault or warning, enter S5; If the fault or warning is not accurately hit, return to S1 and repeat S1-S4 until entering S5; S5: Build the configuration model logic and complete the algorithm model creation system based on configuration technology.

2. The algorithm model creation system based on configuration technology according to claim 1, characterized in that: In S1, the variable nodes include variable nodes and constant nodes, wherein the variable is the data of each device when the train is running, and the constant is a specific value set.

3. The algorithm model creation system based on configuration technology according to claim 1, characterized in that: In S1, the function class node includes a data processing method and a data acquisition module, the data processing method includes arithmetic operations, comparison operations, logical operations, mathematical functions, cutting operations and aggregation functions, the data acquisition module includes a counter node and a timer node, the counter node specifically sets a start condition and an end condition, and collects a group of data according to the set conditions, and the timer node specifically sets a start condition and a collection duration or data volume, and then collects a group of data according to the setting; The custom function class node is specifically: for complex data processing or calculation requirements, when the function class node cannot meet user needs, the user sets parameters and writes Python code to connect the corresponding data to the parameter items of the code, thereby realizing data processing, collection or verification.

4. The algorithm model creation system based on configuration technology according to claim 1, characterized in that: In S1, the end node specifically sets the alarm level and alarm type of the abnormal situation. When the data meets the set abnormal situation, the system reports a fault or warning according to the information set by the end node.

5. The algorithm model creation system based on configuration technology according to claim 4 is characterized in that: Depending on the type of configuration model, the end node sets different data.

6. The algorithm model creation system based on configuration technology according to claim 1, characterized in that: In S2, the connection of nodes is as follows: on the model editing page, drag in variable nodes, function nodes and end nodes in sequence; specifically: The data mapping is achieved by dragging in the variable node. After dragging in the variable node and selecting the corresponding data field, the variable node represents the selected device data itself; then drag in the function node or the custom function node to perform corresponding calculations or collections on the data. When dragging in the function node, select the corresponding calculation logic in the function node, and connect the variable node to the function node to achieve data circulation; when dragging in the custom function node, when connecting the variable node, make the variable node and the parameters set by the custom function correspond one by one; finally, drag in the end node, connect the final calculated result node to the end node, and the end node judges the result data. When the abnormal situation set by the end node is met, the system reports a fault or warning according to the information set by the end node.

7. The algorithm model creation system based on configuration technology according to claim 1, characterized in that: The S3 is specifically: running the combined model logic, and generating test data in a targeted manner according to the combined model logic, and verifying whether the model logic can accurately hit the fault or warning by uploading the test data.

8. The algorithm model creation system based on configuration technology according to claim 1, characterized in that: In S5, the algorithm model creation system based on configuration technology includes configuration model logic, background service logic and model scheduling.

9. The algorithm model creation system based on configuration technology according to claim 8, characterized in that: The background service logic is specifically as follows: according to the edited configuration model logic, take the end node as the starting point, reason forward, record the execution order of all nodes, obtain the corresponding data field according to the variable node, push the data to the node execution order to start calculating the logic of each node, when the data flows to the end node, determine whether the current result meets the threshold set by the end node, and if so, add a new data result.

10. The algorithm model creation system based on configuration technology according to claim 8, characterized in that: The model scheduling is: periodic execution of the configuration model, the periodicity includes real-time scheduling, interval scheduling and single scheduling; The real-time scheduling specifically includes: monitoring data in real time and executing model logic immediately when data is available; The interval scheduling specifically includes: setting a fixed time interval period, within which the configuration model is executed once, and the data content is all the data within the period; The single scheduling specifically means that the configuration model is enabled once every twenty-four hours, and the data content is the data within the twenty-four hours.

Images