阅读说明：本技术 一种基于显示画面直播的ai值守实现系统及方法 (AI (Artificial intelligence) on-duty realization system and method based on live broadcast of display picture ) 是由 钱胜利 于 2020-05-29 设计创作，主要内容包括：本发明提供了一种基于显示画面直播的AI值守实现系统,包括带直播推流编码器的显示器、数据识别区单元、AI视觉库处理单元、AI监控值终端、诊断处理单元；带直播推流编码器的显示器,用于实时监控画面在显示器内的显示面板、云端同步播放；数据识别区单元,无线连接显示器；AI视觉库处理单元,无线连接数据识别区单元；AI监控值终端,输入端无线连接AI视觉库处理单元,诊断处理单元无线连接AI监控值终端。还提供了上述系统的实现方法,在云端完全实现自动闭环监控,能够规避常规依靠人工实时查看监控画面进行判断过程中,可能会出现监控不及时以及判断失误等情况,提高了系统的稳定性,减少了可能的经济损失,增加了安全性。(The invention provides an AI (artificial intelligence) on-duty realization system based on display screen live broadcast, which comprises a display with a live broadcast stream pushing encoder, a data identification area unit, an AI visual library processing unit, an AI monitoring value terminal and a diagnosis processing unit, wherein the data identification area unit is used for identifying the display screen live broadcast stream; the display with the live streaming encoder is used for monitoring synchronous playing of a display panel and a cloud end of a picture in the display in real time; the data identification area unit is wirelessly connected with the display; the AI vision library processing unit is wirelessly connected with the data identification area unit; and the input end of the AI monitoring value terminal is in wireless connection with the AI visual library processing unit, and the diagnosis processing unit is in wireless connection with the AI monitoring value terminal. The system is also provided with an implementation method, automatic closed-loop monitoring is completely realized at the cloud end, the situations that monitoring is not timely and judgment errors and the like may occur in the conventional judgment process by manually checking the monitoring picture in real time can be avoided, the stability of the system is improved, possible economic loss is reduced, and the safety is improved.)

1. An AI on duty implementation system based on live broadcast of display screen, its characterized in that: the system comprises a display with a live broadcast stream pushing encoder, a data identification area unit, an AI visual library processing unit, an AI monitoring value terminal and a diagnosis processing unit; the display with the live streaming encoder is used for monitoring a display panel and a cloud end of a picture in real time and synchronously playing the picture in real time; the data identification area unit receives a transmission signal of a display of the live streaming encoder, and is a data set selected by self-definition under the condition that a display picture is played in real time at a cloud end and is related according to the same monitoring variable; the AI vision library processing unit receives the signal transmitted by the data identification area unit and is used for converting the screenshot of the internal picture of the data identification area unit into monitoring data; the AI monitoring value terminal receives the transmission signal of the AI visual library processing unit and is used for storing and analyzing the monitoring data; the diagnosis processing unit receives the transmission signal of the AI monitoring value terminal and is used for giving a diagnosis result and outputting whether to carry out an alarm signal.

2. The AI attended fulfillment system according to claim 1, wherein: the display with the live streaming encoder comprises a mainboard, a hub interface, a display panel, an encoder and a communication module unit; the main board is wirelessly connected with real-time video signals at the outer end through a hub interface, and two groups of output signal circuits are arranged; one group of output signals are wirelessly connected with the display panel, the other group of output signals are wirelessly connected with the encoder, the encoder is used for compressing the video, and the output signals are wirelessly connected with the communication module unit.

3. An AI on duty implementation method based on display screen live broadcast is characterized in that: use of the AI system according to claim 1 or 2, wherein the specific steps are as follows

(1) Real-time monitoring picture access cloud

Acquiring a real-time monitoring picture, and playing the real-time monitoring picture on a display panel and a cloud end in the display through the display with a live streaming encoder;

(2) collecting data identification area under same monitoring variable

Setting at least one group of monitoring variables, and associating a plurality of groups of data identification areas under the same monitoring variable, wherein the data identification areas are selected and named on a display picture of a cloud end in a self-defined mode;

(3) data processing and AI on duty

And rapidly capturing the images of each group of data identification areas, converting the images and the monitoring values, outputting the converted monitoring data to an AI monitoring value terminal, analyzing the data, and diagnosing.

4. The AI on duty implementation method based on the display live broadcast of claim 3, wherein: in the step (1), the specific operation of synchronously playing the display panel of the display with the live streaming encoder in the display and the cloud is as follows: the display with the live broadcast plug flow encoder is wirelessly connected with the outer end real-time video signal, output signals are output according to two groups of modes, and one group of the output signals is wirelessly communicated with a display panel in the display for real-time display; and the other group is connected to a live streaming encoder through a wireless link, performs video compression, and then outputs the video through a mobile communication module in a 5G or wifi signal transmission mode.

5. The AI on duty implementation method of claim 4 based on the live broadcast of the display screen, wherein: the live streaming encoder is an H.265 encoder, and is an encoder capable of realizing 1080p full high-definition video transmission under the transmission bandwidth lower than 1.5 Mbps.

6. The AI on duty implementation method based on the display live broadcast of claim 3, wherein: in the step (2), under the condition that each data identification area is associated with the same monitoring variable, the corresponding covered numerical values are different.

7. The AI on duty implementation method based on the display live broadcast of claim 3, wherein: in the step (2), the user-defined mode is to select the data identification area according to a mode of setting a plurality of shapes of grabbing frames, wherein the monitoring variable is selected from at least one of voltage, rotating speed, air temperature and equipment temperature of equipment operation, and the monitoring variable is in one-to-one correspondence with the equipment through a name.

8. The AI on duty implementation method based on the display live broadcast of claim 3, wherein: in the step (3), the conversion of the image and the monitoring value is performed by storing the screenshot image in a screenshot area, and connecting an AI vision library processing unit externally through an AI vision library interface in wireless connection.

9. The AI on duty implementation method based on the display live broadcast of claim 3, wherein: the diagnostic process performed in step (3) comprises: classifying the operation state into good, low-efficiency and abnormal alarm modes for alarm processing; or the diagnosis graph of the long-term equipment temperature curve is used as a standard curve, and the real-time data is contrasted to give guidance suggestions for diagnosing and adjusting the temperature interval.

Technical Field

The invention belongs to the field of picture data transmission in the power industry, and particularly relates to an AI (artificial intelligence) on-duty realization system and method based on display picture live broadcast.

Background

At present, in the market, the cloud monitoring equipment often checks the video pictures of real-time monitoring through a camera, and whether the data are normal or not is manual monitoring, so that field personnel are required to check the monitoring pictures in real time and judge the running state, and due to the state factors of people, the situations of untimely monitoring, error judgment and the like can occur, so that certain potential safety hazards and economic loss are caused. Meanwhile, for a video picture monitored in real time, one of video playing factors is influenced, and the stream pushing technology is a technical problem which cannot be ignored in the field.

The push streaming refers to a process of transmitting the content packaged in the acquisition stage to a server, that is, a process of transmitting a live video signal to a network. Aiming at the problems that the requirement of 'push streaming' on a network is high, if the network is unstable, the live broadcasting effect is poor, a viewer can be blocked when watching the live broadcasting, and the watching experience is not smooth. Obviously, the adverse effects caused by the video pictures monitored in real time are amplified, and in severe cases, the serious economic and safety threats are caused.

In addition, in some industrial fields, when a basic network is built, due to the fact that common displays can only independently display real-time pictures of access equipment, if the pictures are shared to a cloud end to be checked in real time, other methods and equipment need to be used again, certain time and energy are needed to be spent for achieving the purpose, the setup time of the network is prolonged, production cost is improved, the whole engineering period is prolonged, and inconvenience is brought.

Disclosure of Invention

In order to solve the technical problem that a monitoring closed loop cannot be formed due to the fact that no reasonable system and method exist when data processing is carried out on real-time videos, the invention provides an AI (artificial intelligence) on-duty implementation system based on display screen live broadcast, and the technical scheme specifically comprises the following steps:

the system comprises a display with a live broadcast stream pushing encoder, a data identification area unit, an AI visual library processing unit, an AI monitoring value terminal and a diagnosis processing unit; the display with the live streaming encoder is used for monitoring a display panel and a cloud end of a picture in real time and synchronously playing the picture in real time; the data identification area unit receives a transmission signal of a display of the live streaming encoder, and is a data set selected by self-definition under the condition that a display picture is played in real time at a cloud end and is related according to the same monitoring variable; the AI vision library processing unit receives the signal transmitted by the data identification area unit and is used for converting the screenshot of the internal picture of the data identification area unit into monitoring data; the AI monitoring value terminal receives the transmission signal of the AI visual library processing unit and is used for storing and analyzing the monitoring data; the diagnosis processing unit receives the transmission signal of the AI monitoring value terminal and is used for giving a diagnosis result and outputting whether to carry out an alarm signal.

As an improvement, the display with the live streaming push encoder comprises a main board, a hub interface, a display panel, an encoder and a communication module unit; the main board is wirelessly connected with real-time video signals at the outer end through a hub interface, and two groups of output signal circuits are arranged; one group of output signals are wirelessly connected with the display panel, the other group of output signals are wirelessly connected with the encoder, the encoder is used for compressing the video, and the output signals are wirelessly connected with the communication module unit.

Meanwhile, a method for realizing AI watch by the system is also provided, and the method comprises the following specific steps:

(1) real-time monitoring picture access cloud

Acquiring a real-time monitoring picture, and playing the real-time monitoring picture on a display panel and a cloud end in the display through the display with a live streaming encoder;

(2) collecting data identification area under same monitoring variable

Setting at least one group of monitoring variables, and associating a plurality of groups of data identification areas under the same monitoring variable, wherein the data identification areas are selected and named on a display picture of a cloud end in a self-defined mode;

(3) data processing and AI on duty

And rapidly capturing the images of each group of data identification areas, converting the images and the monitoring values, outputting the converted monitoring data to an AI monitoring value terminal, analyzing the data, and diagnosing.

As an improvement, in the step (1), the specific operations of the display panel and the cloud side of the display with the live streaming encoder in the display for synchronous playing are as follows: the display with the live streaming push encoder is wirelessly connected with the outer-end real-time video signal, output signals are output in two groups, and one group is electrically communicated with a display panel in the display for real-time display; and the other group is wirelessly connected to a live streaming encoder, video compression is carried out, the video is output to a cloud end through a mobile communication module in a 5G or wifi signal transmission mode, and a display picture is displayed at the cloud end in real time.

As an improvement, the live streaming encoder is an h.265 encoder, and is an encoder capable of realizing 1080p full high-definition video transmission under a transmission bandwidth lower than 1.5 Mbps.

As an improvement, in the step (2), under the condition that each data identification area is associated with the same monitoring variable, the corresponding covered numerical values are different.

As an improvement, in the step (2), the data identification area is selected in a self-defining mode according to a mode of setting a plurality of shapes of grabbing frames, wherein the monitoring variable is selected from at least one of voltage, rotating speed, air temperature and equipment temperature of equipment operation, and the monitoring variable is in one-to-one correspondence with the equipment through a name.

In the step (3), the image and the monitored value are converted by storing the screenshot image in the screenshot area and connecting an AI visual library processing unit externally through an AI visual library interface in wireless connection.

As a refinement, the diagnostic process performed in step (3) includes: classifying the operation state into good, low-efficiency and abnormal alarm modes for alarm processing; or the diagnosis graph of the long-term equipment temperature curve is used as a standard curve, and the real-time data is contrasted to give guidance suggestions for diagnosing and adjusting the temperature interval.

Has the advantages that: the display picture live broadcast-based AI on-duty implementation system provided by the invention is a system which is based on an AI intelligent on-duty technology and can completely implement automatic closed-loop monitoring at the cloud, thereby avoiding the situations of monitoring untimely and judgment error and the like which can occur in the conventional process of judging the possible situations by manually checking the monitoring picture in real time, improving the stability of the system, reducing the possible economic loss and increasing the safety.

In addition, through setting up the display of taking live push stream encoder, real-time broadcast is carried out at display panel, high in the clouds in the display, can solve current display and can not realize showing real-time picture when, the high in the clouds looks over the technical problem of picture in real time, has realized in the thing networking of industrial field upgrading and the process of building, consuming time long, consume the shortcoming that the resource is many.

Drawings

Fig. 1 is a schematic diagram of the principle of the present invention.

Fig. 2 is a schematic diagram of a display with a live push stream encoder according to the present invention.

Detailed Description

The figures of the present invention are further described below in conjunction with the embodiments.

An AI on duty implementation system based on live broadcast of display frame specifically is: the system comprises a display with a live broadcast stream pushing encoder, a data identification area unit, an AI visual library processing unit, an AI monitoring value terminal and a diagnosis processing unit; the display with the live streaming encoder is used for monitoring a display panel and a cloud end of a picture in real time and synchronously playing the picture in real time; the data identification area unit receives a transmission signal of a display of the live streaming encoder, and is a data set selected by self-definition under the condition that a display picture is played in real time at a cloud end and is related according to the same monitoring variable; the AI vision library processing unit receives the signal transmitted by the data identification area unit and is used for converting the screenshot of the internal picture of the data identification area unit into monitoring data; the AI monitoring value terminal receives the transmission signal of the AI visual library processing unit and is used for storing and analyzing the monitoring data; the diagnosis processing unit receives the transmission signal of the AI monitoring value terminal and is used for giving a diagnosis result and outputting whether to carry out an alarm signal.

As a specific implementation mode of the invention, the display with the live streaming encoder comprises a mainboard, a hub interface, a display panel, an encoder and a communication module unit; the main board is wirelessly connected with real-time video signals at the outer end through a hub interface, and two groups of output signal circuits are arranged; one group of output signals are wirelessly connected with the display panel, the other group of output signals are wirelessly connected with the encoder, the encoder is used for compressing the video, and the output signals are wirelessly connected with the communication module unit.

Selectively, the output signal of the communication module unit is wirelessly connected to a cloud end, and the cloud end is used for viewing the display picture in real time.

The hub interface is an interface passing through a VGA hub or an HDMI hub, the encoder is an H.265 encoder, the encoder can realize 1080p full high-definition video transmission under the transmission bandwidth lower than 1.5Mbps, the communication module unit is a mobile communication module unit in a 5G or wifi signal transmission mode, and transmits signals to the cloud end through an RTMP protocol, wherein the signal transmission mode is not limited to 5G or wifi, only can realize the same function of the invention, and can be selected as other signal transmission modes in the invention, such as 4G, 3G, Bluetooth and the like, which belong to the protection scope of the invention.

Preferably, encoder electric connection has control switch, and the switch can be controlled, through opening, closing of adjustment control switch, further ground electric connection high in the clouds picture, controls opening, closing of high in the clouds picture, and is simple, convenient, swift.

The invention also provides a method for implementing AI watch by the system, which comprises the following steps:

(1) real-time monitoring picture access cloud

Acquiring a real-time monitoring picture, and playing the real-time monitoring picture on a display panel and a cloud end in the display through the display with a live streaming encoder;

(2) collecting data identification area under same monitoring variable

Setting at least one group of monitoring variables, and associating a plurality of groups of data identification areas under the same monitoring variable, wherein the data identification areas are selected and named on a display picture of a cloud end in a self-defined mode;

(3) data processing and AI on duty

And rapidly capturing the images of each group of data identification areas, converting the images and the monitoring values, outputting the converted monitoring data to an AI monitoring value terminal, analyzing the data, and diagnosing.

In the step (1), the specific operation of synchronously playing the display panel of the display with the live streaming encoder in the display and the cloud is as follows: the display with the live streaming push encoder is wirelessly connected with the outer-end real-time video signal, output signals are output in two groups, and one group is electrically communicated with a display panel in the display for real-time display; and the other group is wirelessly connected to a live streaming encoder, video compression is carried out, the video is output to a cloud end through a mobile communication module in a 5G or wifi signal transmission mode, and a display picture is displayed at the cloud end in real time.

In the step (2), under the condition that each data identification area is associated with the same monitoring variable, the corresponding covered numerical values are different. The user-defined mode is to select the data identification area according to a mode of setting a plurality of shapes of grabbing frames, wherein the monitoring variable is selected to be at least one of the voltage, the rotating speed, the air temperature and the equipment temperature of the equipment, and the monitoring variable is in one-to-one correspondence with the equipment through the name.

In the step (3), the conversion of the image and the monitoring value is performed by storing the screenshot image in a screenshot area, and connecting an AI vision library processing unit externally through an AI vision library interface in wireless connection. The AI visual library processing units are Baidu, Tencent, little ant, etc., but are not limited to just these.

The diagnostic process performed in step (3) comprises: classifying the operation state into good, low-efficiency and abnormal alarm modes for alarm processing; or the diagnosis graph of the long-term equipment temperature curve is used as a standard curve, and the real-time data is contrasted to give guidance suggestions for diagnosing and adjusting the temperature interval.