Distributed signal lamp fault detection system based on intelligent fault diagnosis

Abstract

The invention relates to a distributed signal lamp fault detection system based on intelligent fault diagnosis. The system comprises a computing power machine, a metering machine and a central platform. The computing power machine receives data from the metering machine, interacts with the central platform through wireless communication, and reports fault information. The metering machine is installed on a signal lamp, collects voltage, current and attitude angle parameters, and sends the parameters to the computing power machine through low-power-consumption Bluetooth or Wi-Fi. The computing power machine comprises a first microcontroller and a computing power chip and is used for receiving and integrating information reported by the metering machine and carrying out model operation through various intelligent fault diagnosis algorithms. And the central platform receives fault information transmitted by the computing power machine and sends a maintenance instruction to the operation and maintenance personnel PDA through the mobile communication network. The computing power machine can dynamically adjust the used fault diagnosis algorithm according to different application scenes. According to the method, the accuracy and reliability of fault detection are remarkably improved, the operation and maintenance cost is reduced, the operation and maintenance efficiency is improved, and the safety and smoothness of urban traffic are ensured.

Description

Technical Field

[0001] The present invention relates to the technical field of traffic signal control, and in particular to a distributed signal light fault detection system based on intelligent fault diagnosis. Background Art

[0002] Traffic lights are the core equipment for controlling and regulating key nodes in the urban traffic organization and management system, and are the key to the smooth operation of the entire road network. With the increase in the number of signal-controlled intersections, abnormal traffic congestion caused by traffic light failures is also gradually increasing, which brings certain safety hazards.

[0003] The normal operation of traffic lights is a prerequisite for the normal operation of road network traffic. Timely discovery and effective repair of faulty traffic lights is a necessary condition to ensure the normal operation of the road network. Whether the normal and stable operation of urban traffic lights is directly related to the safety of urban traffic.

[0004] At present, there are several main methods for traffic light operation and maintenance:

[0005] (1) Manual inspection / reporting by citizens;

[0006] (2) The fault detection function of the signal machine board;

[0007] (3) Reuse existing electronic police surveillance for video recognition;

[0008] (4) A fault detector independently installed on the outside of the signal light.

[0009] Manual inspection / citizen reporting methods often have problems such as low investigation rate, high cost, and low efficiency. The fault detection function of the signal machine board is generally based on a dedicated network, which is not convenient for quickly feeding back fault information to the traffic police and emergency repair units. The accuracy of the scheme of multiplexing electric police for visual recognition inspection is difficult to guarantee. Due to factors such as the angle of the electric police and the control of the signal machine, false alarms are likely to occur. There are two schemes for installing signal light fault detectors independent of existing products. One is a unit-level signal fault detector. This device achieves full coverage of all faults by connecting the fault detection device to each signal light. The other is a signal machine-level signal light fault detection device. This device is installed between the signal machine and the signal light, and determines whether the signal light has a fault by detecting the current and voltage of each signal light. This type of equipment currently only makes judgments through a simple threshold setting method. False alarms often occur in the field due to changes in the timing plan and other reasons. The detection accuracy is not high and needs to be further improved. Summary of the invention

[0010] To solve the above technical problems, the present invention provides a distributed signal light fault detection system based on intelligent fault diagnosis, which aims to improve the accuracy of signal light fault detection, enrich the scope of signal light fault detection, and ensure the high availability of signal lights.

[0011] A distributed signal light fault detection system based on intelligent fault diagnosis of the present invention comprises:

[0012] (1) The computing machine is used to receive data from the metering machine, interact with the central platform through wireless communication, and report fault information;

[0013] (2) A metering machine, installed on each signal light fixture, is used to collect the voltage parameters, current parameters and attitude angle parameters of the signal light and send them to the computing machine wirelessly;

[0014] (3) The central platform is used to receive fault information from the computing machine and send maintenance instructions to the PDA of the operation and maintenance personnel through the network. PDA refers to a personal digital assistant, which is an electronic device.

[0015] Preferably, the computing machine comprises:

[0016] (1) A first microcontroller (MCU), configured to receive information reported by each metering machine through a first wireless module, integrate and summarize the information, and transmit the information to a computing chip for model calculation;

[0017] (2) Computing chips are used to run multiple algorithms to determine whether traffic lights are faulty.

[0018] Preferably, the metering machine comprises:

[0019] (1) a second microcontroller (MCU), configured to measure the voltage and current of the driving signal lamp through each metering module, wherein the metering machine includes a plurality of metering modules;

[0020] (2) Gyroscope, used to detect the attitude angle of the signal light;

[0021] (3) The second wireless module is used to report the summarized information to the computing machine.

[0022] Preferably, the central platform can send emergency repair instructions to the PDA of the operation and maintenance personnel through the network, and can receive feedback information from the operation and maintenance personnel to facilitate rapid response and processing of traffic light failures.

[0023] A signal light fault detection method based on intelligent fault diagnosis is implemented by applying the distributed signal light fault detection system based on intelligent fault diagnosis, and specifically comprises the following steps:

[0024] S1: Detect the voltage, current and attitude angle information of each signal light through a measuring machine;

[0025] S2: wirelessly upload the detected information to the computing machine;

[0026] S3: The computing machine receives and integrates the information reported by the metering machine, performs model calculations through the computing chip, and determines the fault status of the signal light;

[0027] S4: Report the identified fault information to the central platform;

[0028] S5: The central platform receives the fault information and sends the emergency repair instructions to the operation and maintenance personnel's PDA through the network.

[0029] Preferably, an algorithm based on intelligent fault diagnosis is run to automatically identify the health status of the signal light.

[0030] Preferably, the intelligent fault diagnosis algorithm includes:

[0031] Machine learning theories such as artificial neural network (ANN), support vector machine (SVM), deep neural network (DNN), etc.

[0032] Preferably, low-power Bluetooth or Wi-Fi is used for data transmission between the metering machine and the computing machine.

[0033] Preferably, the communication between the central platform and the PDA of the operation and maintenance personnel adopts the mobile communication network to ensure the real-time transmission of information.

[0034] Preferably, the computing machine can dynamically adjust the fault diagnosis algorithm used according to different application scenarios to improve the flexibility and accuracy of fault detection.

[0035] Compared with the prior art, the present invention has the following beneficial effects:

[0036] The present invention significantly improves the accuracy and reliability of fault detection through a distributed signal light fault detection system based on intelligent fault diagnosis. Specifically:

[0037] Improve detection accuracy: By using intelligent fault diagnosis algorithms (such as ANN, SVM, DNN), the system can automatically identify the health status of traffic lights, reducing false positives and missed negatives.

[0038] Expanded detection range: The metering machine not only detects voltage and current, but also multi-dimensional data such as attitude angle, and can detect more types of faults, such as tilted or damaged traffic lights.

[0039] Improve system reliability: The distributed architecture ensures that even if a meter fails, it will not affect the detection of other signal lights. The system supports dynamic adjustment algorithms to adapt to different scenarios and is easy to upgrade and maintain.

[0040] Reduce costs: The system uses low-power communication and configures metering machines on demand, reducing unnecessary costs and complexity.

[0041] Improve operation and maintenance efficiency: The central platform receives fault information in real time and sends emergency repair instructions through the mobile communication network, achieving rapid response. Automated diagnosis reduces the workload of manual inspections and improves the efficiency of fault handling.

[0042] Through these improvements, the present invention has important application value in urban traffic management and effectively ensures the safety and smoothness of urban traffic. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] Figure 1 It is a schematic diagram of the framework of the signal light fault detection system based on intelligent fault diagnosis of the present invention;

[0044] Figure 2 It is a schematic diagram of the composition of the metering machine of the present invention;

[0045] Figure 3 It is a schematic diagram of the composition of the computing machine of the present invention; DETAILED DESCRIPTION

[0046] The following embodiments of the present invention are described in further detail in conjunction with the accompanying drawings and examples. The following embodiments are used to illustrate the present invention but are not intended to limit the scope of the present invention. Figures 1 to 3 As shown,

[0047] Example 1: Basic Configuration

[0048] Application scenario: small intersections with fewer traffic lights and relatively simple requirements for fault detection.

[0049] System composition:

[0050] Computing machine: It is placed inside the signal machine and is responsible for receiving data from each metering machine, interacting with the central platform through wireless communication, and reporting fault information. The computing machine includes a first MCU and a computing chip, wherein the first MCU receives the information reported by each metering machine through the first wireless module, integrates and summarizes it, and transmits it to the computing chip for model calculation. The computing chip runs multiple sets of algorithms to determine whether the traffic light is faulty.

[0051] Metering machine: Installed on the housing of each signal light, it is responsible for detecting the voltage, current, attitude angle and other information of each signal light, and sending it to the computing machine wirelessly. The metering machine includes a second MCU, a gyroscope and a second wireless module. The second MCU measures the voltage and current of the driving signal light through each metering module, and the gyroscope detects the attitude angle of the signal light, which is summarized and reported to the computing machine through the second wireless module.

[0052] Central platform: Receives fault information from the computing machine and sends maintenance instructions to the PDA of the operation and maintenance personnel through the network. PDA refers to personal digital assistant, which is an electronic device.

[0053] Workflow:

[0054] Data collection: Each metering machine detects the voltage, current and attitude angle information of the signal light.

[0055] Data transmission: The metering machine sends the detected information to the computing machine via low-power Bluetooth or Wi-Fi.

[0056] Data processing: The computing machine receives and integrates the information reported by the metering machine, performs model calculations through the computing chip, and determines the fault status of the traffic light.

[0057] Fault report: The computing machine will report the fault information it has identified to the central platform.

[0058] Repair instructions: The central platform receives fault information and sends repair instructions to the operation and maintenance personnel's PDA via the mobile communication network.

[0059] Example 2: Medium-sized intersection

[0060] Application scenario: Medium-sized intersections with a large number of traffic lights have high requirements for fault detection and require higher detection accuracy and reliability.

[0061] System composition:

[0062] Computing machine: In addition to the basic configuration, the computing machine supports more complex deep learning models, such as deep neural networks (DNN). The computing machine is also equipped with high-performance computing chips to support large-scale data processing and real-time computing. The computing machine can update the algorithm model through the cloud to maintain the latest fault diagnosis capabilities.

[0063] Metering machine: In addition to the basic configuration, the metering machine can be equipped with more sensors, such as temperature and humidity sensors, light sensors, etc., to provide more dimensional fault detection data. The second wireless module of the metering machine supports multiple wireless communication protocols to ensure the reliability and stability of data transmission.

[0064] Central platform: The central platform integrates big data analysis and cloud computing technologies, can process large amounts of data in real time, and continuously optimize fault diagnosis models through machine learning algorithms. The central platform can also be connected to the city management platform to achieve real-time sharing and collaborative processing of fault information.

[0065] Workflow:

[0066] Data collection: Each metering machine detects the voltage, current, attitude angle, temperature, humidity, light and other information of the signal light.

[0067] Data transmission: The metering machine sends the detected information to the computing machine via low-power Bluetooth or Wi-Fi.

[0068] Data processing: The computing machine receives and integrates the information reported by the metering machine, performs model calculations through the computing chip, and determines the fault status of the traffic light. The computing machine supports more complex deep learning models, such as DNN, and can dynamically adjust the fault diagnosis algorithm used according to different application scenarios.

[0069] Fault report: The computing machine will report the fault information it has identified to the central platform.

[0070] Repair instructions: The central platform receives fault information and sends repair instructions to the PDA of the operation and maintenance personnel through the mobile communication network. The central platform can also be connected with the city management platform to achieve real-time sharing and collaborative processing of fault information.

[0071] Example 3: Large intersection

[0072] Application scenarios: Large intersections with a large number of traffic lights have extremely high requirements for fault detection, requiring a high degree of automation and intelligence.

[0073] System composition:

[0074] Computing machine: In addition to the configuration for medium-sized intersections, the computing machine supports more advanced algorithm models such as reinforcement learning and transfer learning. The computing machine is also equipped with a high-performance GPU to support large-scale data processing and real-time computing. The computing machine can update the algorithm model in real time through the cloud to maintain the latest fault diagnosis capabilities.

[0075] Metering machine: In addition to the configuration of medium-sized intersections, the metering machine can be equipped with more sensors, such as vibration sensors, sound sensors, etc., to provide more comprehensive fault detection data. The second wireless module of the metering machine supports a variety of wireless communication protocols to ensure the reliability and stability of data transmission.

[0076] Central platform: The central platform integrates big data analysis, cloud computing and edge computing technologies, can process large amounts of data in real time, and continuously optimize fault diagnosis models through machine learning algorithms. The central platform can also connect with multiple city management platforms to achieve real-time sharing and collaborative processing of fault information.

[0077] Workflow:

[0078] Data collection: Each metering machine detects the voltage, current, attitude angle, temperature and humidity, light, vibration, sound and other information of the signal light.

[0079] Data transmission: The metering machine sends the detected information to the computing machine via low-power Bluetooth or Wi-Fi.

[0080] Data processing: The computing machine receives and integrates the information reported by the metering machine, performs model calculations through the computing chip, and determines the fault status of the traffic light. The computing machine supports more advanced algorithm models, such as reinforcement learning and transfer learning, and can dynamically adjust the fault diagnosis algorithm used according to different application scenarios.

[0081] Fault report: The computing machine will report the fault information it has identified to the central platform.

[0082] Repair instructions: The central platform receives fault information and sends repair instructions to the PDA of the operation and maintenance personnel through the mobile communication network. The central platform can also connect with multiple city management platforms to achieve real-time sharing and collaborative processing of fault information.

[0083] The computing machine of the present invention receives and integrates the reported information of the metering machine, performs model calculation through the computing chip, and determines the specific process of the fault state of the signal light as follows:

[0084] 1. Data reception and preprocessing

[0085] Data reception:

[0086] Metering machine reports data: Each metering machine sends the detected signal light voltage, current, attitude angle and other information to the computing machine through the second wireless module.

[0087] Data format: The reported data is usually sent in a structured format, such as JSON format, containing the readings and timestamps of each sensor.

[0088] Data preprocessing:

[0089] Data analysis: The first MCU of the computing machine receives data through the first wireless module, analyzes the reported data packets, and extracts the readings of each sensor.

[0090] Data cleaning: Remove invalid or abnormal data points, such as voltage or current values ​​that are outside a reasonable range, or obviously erroneous attitude angle readings.

[0091] Data normalization: Normalize the readings of different sensors so that they can be compared and processed in the same dimension. For example, normalize voltage and current values ​​to between 0 and 1.

[0092] 2. Data integration and feature extraction

[0093] Data Integration:

[0094] Time alignment: Align the data from different sensors at the same time point to form a complete data record.

[0095] Data aggregation: Aggregate data collected multiple times in the same time period, such as taking the average or median, to reduce the impact of noise.

[0096] Feature extraction:

[0097] Basic features: Extract basic features such as voltage, current, and attitude angle.

[0098] Advanced features: Calculate advanced features based on basic features, such as voltage fluctuation rate, current change rate, attitude angle change rate, etc. These advanced features can help the model capture the operating status of the traffic light more accurately.

[0099] Historical features: Combine historical data to extract trend features, such as the historical change trend of voltage and current, the stability of attitude angle, etc.

[0100] 3. Model calculation

[0101] Model selection:

[0102] Multi-model support: The computing machine supports a variety of fault diagnosis algorithms, including artificial neural networks (ANN), support vector machines (SVM), deep neural networks (DNN), etc.

[0103] Dynamic adjustment: Dynamically select the appropriate model for operation according to different application scenarios and fault types. For example, for common electrical faults, you can choose the SVM model; for complex multi-factor faults, you can choose the DNN model.

[0104] Model input:

[0105] Feature vector: The extracted features are combined into a feature vector as the input of the model.

[0106] Data format: A feature vector is usually a multidimensional array, with each dimension corresponding to a feature.

[0107] Model operation:

[0108] Forward propagation: Input the feature vector into the selected model for forward propagation calculation. For example, for a DNN model, the input layer passes the feature vector to the hidden layer, the hidden layer performs nonlinear transformation through the activation function, and finally the output layer produces the prediction result.

[0109] Fault classification: The model outputs one or more probability values, representing the probability of the signal light being in different fault states. For example, the output may include the probabilities of being in "normal", "electrical fault", "physical fault", etc.

[0110] 4. Fault diagnosis and decision making

[0111] Threshold determination:

[0112] Set threshold: Set the probability threshold of each fault state based on experience or experimental data. For example, if the probability of "electrical fault" is greater than 0.8, it is judged that the traffic light has an electrical fault.

[0113] Multi-model fusion: If multiple models are used, the output results of multiple models can be combined through a voting mechanism or weighted average to improve the accuracy of judgment.

[0114] Fault confirmation:

[0115] Continuous monitoring: If it is preliminarily determined that the signal light is faulty, continue to monitor the data for a period of time to confirm the persistence and severity of the fault.

[0116] Fault type refinement: Based on the specific characteristic values, the fault type can be further refined. For example, electrical faults may be subdivided into different types such as short circuit, open circuit, overload, etc.

[0117] Fault Report:

[0118] Generate report: Generate a fault report with the final determined fault type, fault time, fault location and other information.

[0119] Report to the central platform: Send the fault report to the central platform via the network for further processing and repair.

[0120] 5. Feedback and Optimization

[0121] Feedback Mechanism:

[0122] Real-time feedback: After receiving the fault report, the central platform immediately sends repair instructions to the operation and maintenance personnel's PDA to achieve rapid response.

[0123] Historical data recording: Record the results of each fault detection to form historical data.

[0124] Model optimization:

[0125] Online learning: Based on historical data and actual maintenance results, the model is regularly updated and optimized to improve the accuracy and robustness of fault detection.

[0126] Offline training: In a laboratory environment, a large amount of labeled data is generated by simulating various fault types for training and validating the model.

[0127] The computing machine of the present invention dynamically adjusts the fault diagnosis algorithm used according to different application scenarios. The dynamic adjustment process is specifically as follows:

[0128] Application scenario analysis

[0129] Collect data: First, the computing machine needs to collect relevant data about the current application scenario, such as ambient temperature, humidity, traffic flow, frequency of traffic light use, etc.

[0130] Define criteria: Determine which factors are critical in influencing the choice of algorithm, for example, in extreme weather conditions, a more sensitive fault detection algorithm may be required to ensure safety.

[0131] 2. Algorithm library preparation

[0132] Algorithm library construction: Pre-install a variety of fault diagnosis algorithms, such as artificial neural network (ANN), support vector machine (SVM), deep neural network (DNN), etc., to form an algorithm library.

[0133] Algorithm feature record: Detailed record of the applicable conditions, advantages, disadvantages, etc. of each algorithm to provide a basis for subsequent algorithm selection.

[0134] 3. Dynamic evaluation

[0135] Performance evaluation: Regularly or under specific conditions, the performance of the currently used algorithm is evaluated, such as accuracy, response time, and other indicators.

[0136] Demand matching: Match the current application scenario requirements with the characteristics of each algorithm in the algorithm library to find the algorithm that best suits the current situation.

[0137] 4. Algorithm Switching

[0138] Decision making: Based on the evaluation results and demand matching, decide whether to switch algorithms.

[0139] Smooth transition: If you decide to switch algorithms, you need to design a mechanism to ensure service continuity and stability during the algorithm switching process to avoid service interruptions or false detections caused by the switch.

[0140] Parameter adjustment: For a newly selected algorithm, you may need to adjust relevant parameters according to the current application scenario to optimize its performance.

[0141] 5. Continuous Optimization

[0142] Feedback loop: After the algorithm is switched, continue to monitor its performance and collect feedback from operation and maintenance personnel for subsequent algorithm optimization or adjustment.

[0143] Learning mechanism: Introducing a machine learning mechanism allows the system to learn from each adjustment and gradually optimize the algorithm selection strategy.

[0144] 6. Security

[0145] Anomaly detection: During the entire process, a complete anomaly detection mechanism is required. Once an anomaly is detected during algorithm switching or operation, immediate measures should be taken to restore or alarm.

[0146] Backup plan: Ensure that even in the worst case scenario, there are backup algorithms available to maintain the normal operation of the system.

[0147] The above is only a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the technical principles of the present invention. These improvements and modifications should also be regarded as the scope of protection of the present invention.

Claims

1. A distributed signal light fault detection system based on intelligent fault diagnosis, characterized in that: include: (1) The computing machine is used to receive data from the metering machine, interact with the central platform through wireless communication, and report fault information; (2) A metering machine, installed on each signal light fixture, is used to collect the voltage parameters, current parameters and attitude angle parameters of the signal light and send them to the computing machine wirelessly; (3) The central platform is used to receive fault information from the computing machine and send maintenance instructions to the PDA of the operation and maintenance personnel.

2. A distributed signal light fault detection system based on intelligent fault diagnosis as claimed in claim 1, characterized in that: The computing machine comprises: (1) A first microcontroller, configured to receive information reported by each metering machine through a first wireless module, integrate and aggregate the information, and transmit the information to a computing chip for model calculation; (2) Computing chip, used to run model operations to determine whether the traffic light is faulty.

3. A distributed signal light fault detection system based on intelligent fault diagnosis as claimed in claim 2, characterized in that: The metering machine comprises: (1) A second microcontroller, used to measure the voltage and current of the driving signal lamp through each metering module, and the metering machine includes a plurality of metering modules; (2) Gyroscope, used to detect the attitude angle of the signal light; (3) The second wireless module is used to report the summarized information to the computing machine.

4. A distributed signal light fault detection system based on intelligent fault diagnosis as claimed in claim 1, characterized in that: The central platform can send emergency repair instructions to the PDA of the operation and maintenance personnel through the network, and can receive feedback information from the operation and maintenance personnel to facilitate rapid response and processing of traffic light failures.

5. A signal light fault detection method based on intelligent fault diagnosis as claimed in claim 1, characterized in that: Low-power Bluetooth or Wi-Fi is used for data transmission between the metering machine and the computing machine.

6. A signal light fault detection method based on intelligent fault diagnosis as claimed in claim 4, characterized in that: The communication between the central platform and the PDA of the operation and maintenance personnel adopts the mobile communication network to ensure the real-time transmission of information.

7. A signal light fault detection method based on intelligent fault diagnosis, characterized in that: The method is implemented by using a distributed signal light fault detection system based on intelligent fault diagnosis as described in any one of claims 1 to 6, specifically comprising the following steps: S1: Detect the voltage, current and attitude angle information of each signal light through a measuring machine; S2: wirelessly upload the detected information to the computing machine; S3: The computing machine receives and integrates the information reported by the metering machine, performs model calculations through the computing chip, and determines the fault status of the signal light; S4: Report the identified fault information to the central platform; S5: The central platform receives the fault information and sends the emergency repair instructions to the operation and maintenance personnel's PDA through the network.

8. A signal light fault detection method based on intelligent fault diagnosis as claimed in claim 7, characterized in that: Run model operations based on intelligent fault diagnosis to automatically identify the health status of traffic lights.

9. A signal light fault detection method based on intelligent fault diagnosis as claimed in claim 8, characterized in that: The intelligent fault diagnosis algorithm includes: Artificial neural network algorithm, support vector machine algorithm and deep neural network algorithm.

10. A signal light fault detection method based on intelligent fault diagnosis as claimed in claim 9, characterized in that: The computing machine dynamically adjusts the fault diagnosis algorithm used according to different application scenarios to improve the flexibility and accuracy of fault detection.

Images