Disclosure of Invention
The present invention aims to solve at least one of the technical problems existing in the prior art; therefore, the invention provides an equipment appearance detection system based on an image recognition technology and an infrared thermal imaging technology, which is used for solving the technical problem that in the prior art, in the process of inspecting a high-voltage direct-current converter station, fine anomalies of equipment in the long-term use process are difficult to find, so that the state of the high-voltage direct-current converter station cannot be accurately evaluated.
In order to achieve the above object, a first aspect of the present invention provides an apparatus appearance detection system based on an image recognition technology and an infrared thermal imaging technology, which includes a central control module, and an intelligent terminal and a plurality of data acquisition modules connected with the central control module; the intelligent terminal is used for fault monitoring and early warning;
the data acquisition module acquires infrared image data with long time sequence through an infrared camera connected with the data acquisition module, and sends the infrared image data to the central control module after preprocessing;
the central control module extracts time sequence temperature characteristics from the infrared image data with long time sequence, analyzes the time sequence temperature characteristics by combining a curve fitting method, and obtains a temperature risk score according to an analysis result;
the central control module collects equipment appearance images based on the equipment risk scores, analyzes the equipment appearance through an image recognition technology after correcting the equipment appearance images, and obtains the equipment appearance scores according to analysis results.
Preferably, the central control module is respectively in communication and/or electric connection with the intelligent terminal and the plurality of data acquisition modules; the data acquisition modules are in one-to-one correspondence with the equipment;
the data acquisition module is respectively communicated and/or electrically connected with the infrared camera and the high-definition camera; the infrared camera is used for collecting infrared image data, and the high-definition camera is used for collecting equipment appearance images.
Preferably, the central control module extracts time-series temperature characteristics based on long-time-series infrared image data, and the central control module comprises:
continuously acquiring infrared image data of the equipment in the working process by an infrared camera;
and extracting abnormal temperature data of the surface of the equipment from the infrared image data, and splicing average temperature data of a plurality of acquisition moments of the positions corresponding to the abnormal temperature data to generate time sequence temperature characteristics.
Preferably, the central control module analyzes time sequence temperature characteristics in combination with a curve fitting method, and comprises the following steps:
fitting a plurality of average temperature data in the time sequence temperature characteristic based on a curve fitting method to obtain a temperature characteristic curve;
identifying discrete features and trend features of the time sequence temperature features through a temperature feature curve; wherein the discrete feature represents a discrete state of the time-series temperature feature, and the trend feature represents a trend of change of the time-series temperature feature.
Preferably, calculating a temperature wind direction score corresponding to the time sequence temperature characteristic through the temperature characteristic curve includes:
setting a temperature characteristic threshold value, and marking a rectangular area surrounded by the temperature characteristic threshold value in a coordinate axis as a standard area; counting the proportion of the temperature characteristic curve in the standard region, and marking the proportion as a discrete characteristic;
calculating the ratio of the mean value of the first derivatives of the three acquisition moments at the end of the temperature characteristic curve to the mean value of the first derivatives of all the acquisition moments, and marking the ratio as a trend characteristic.
Preferably, calculating a temperature risk score from the discrete features and the trend features comprises:
marking the discrete features as LST and the trend features as QST;
obtaining a temperature risk score WFP by the formula wfp=α×|qst-qsy|/LST; where α is a scaling factor greater than 0 and QSY is a trend feature threshold.
Preferably, when the temperature risk score is greater than the temperature risk threshold, marking the position corresponding to the temperature risk score as an abnormal position; the central control module collects equipment appearance images through the high-definition camera, identifies material strain and fine deformation at abnormal positions based on the equipment appearance images, and calculates equipment appearance scores according to the material strain and the fine deformation.
Preferably, the obtaining of the equipment appearance score from material strain and fine deformation comprises:
identifying whether material strain or fine deformation occurs at the abnormal position through an image identification technology; if yes, marking the risk position of the abnormal position; if not, not processing;
counting the duty ratio of the corresponding areas of the plurality of risk positions on the appearance of the equipment, and obtaining the appearance score of the equipment according to the occupied ratio; wherein, the larger the risk location area, the smaller the equipment appearance score.
Compared with the prior art, the invention has the beneficial effects that: according to the invention, long-time sequence infrared image data of the equipment are acquired in real time through the infrared camera, abnormal positions which possibly have faults are judged through analysis of the infrared image data, and then a plurality of abnormal positions are identified and counted according to the appearance images of the equipment, so that the appearance scores of the equipment are obtained. According to the invention, the running state of the equipment is analyzed through the infrared image data, the position where the abnormality possibly occurs can be more sensitively identified, and then the overall analysis is carried out by combining the appearance image of the equipment, so that the equipment can be objectively and accurately evaluated.
Detailed Description
The technical solutions of the present invention will be clearly and completely described in connection with the embodiments, and it is obvious that the described embodiments are only some embodiments of the present invention, not all embodiments. All other embodiments, which can be made by those skilled in the art based on the embodiments of the invention without making any inventive effort, are intended to be within the scope of the invention.
Referring to fig. 1, an embodiment of the first aspect of the present invention provides an apparatus appearance detection system based on an image recognition technology and an infrared thermal imaging technology, which includes a central control module, and an intelligent terminal and a plurality of data acquisition modules connected with the central control module; the intelligent terminal is used for fault monitoring and early warning; the data acquisition module acquires infrared image data with long time sequence through an infrared camera connected with the data acquisition module, and sends the infrared image data to the central control module after preprocessing; the central control module extracts time sequence temperature characteristics from the infrared image data with long time sequence, analyzes the time sequence temperature characteristics by combining a curve fitting method, and obtains a temperature risk score according to an analysis result; the central control module collects equipment appearance images based on the equipment risk scores, analyzes the equipment appearance through an image recognition technology after correcting the equipment appearance images, and obtains the equipment appearance scores according to analysis results.
In the prior art, when the high-voltage direct-current converter station is inspected, whether the high-voltage direct-current converter station can normally work or not and whether all working parameters are within a set range or not are generally checked, inspection work is also regularly performed by workers, fine anomalies of the high-voltage direct-current converter station cannot be accurately identified in time, and faults cannot be pre-judged.
According to the invention, long-time sequence infrared image data of the equipment are acquired in real time through the infrared camera, abnormal positions which possibly have faults are judged through analysis of the infrared image data, and then a plurality of abnormal positions are identified and counted according to the appearance images of the equipment, so that the appearance scores of the equipment are obtained. According to the invention, the running state of the equipment is analyzed through the infrared image data, the position where the abnormality possibly occurs can be more sensitively identified, and then the overall analysis is carried out by combining the appearance image of the equipment, so that the equipment can be objectively and accurately evaluated.
In the invention, a central control module is respectively communicated and/or electrically connected with an intelligent terminal and a plurality of data acquisition modules; the data acquisition modules are in one-to-one correspondence with the equipment; the data acquisition module is respectively communicated and/or electrically connected with the infrared camera and the high-definition camera.
The appearance detection system of the whole equipment is essentially that the infrared image data is used for judging which positions of the high-voltage direct-current converter station have abnormal temperature data, and then the high-definition camera is used for shooting the appearance image of the equipment to identify the material performance and the fine deformation. Therefore, an infrared camera and a high-definition camera are correspondingly arranged for each device, the infrared camera and the high-definition camera are connected with the same data acquisition module, and the data acquisition module is responsible for image data acquisition and image preprocessing of the corresponding device. The central control module is connected with a plurality of data acquisition modules, namely, a plurality of devices can be detected at the same time. The intelligent terminal is mainly used for checking the states of all the devices and detecting the states.
The infrared camera is used for collecting infrared image data, and the high-definition camera is used for collecting equipment appearance images. Whether the infrared camera or the high-definition camera can acquire all-dimensional image data of the equipment, the infrared camera works regularly in the whole detection process, and the high-definition camera determines whether to work according to the analysis result of the infrared image data.
The central control module extracts time sequence temperature characteristics based on long time sequence infrared image data, and comprises the following steps: continuously acquiring infrared image data of the equipment in the working process by an infrared camera; and extracting abnormal temperature data of the surface of the equipment from the infrared image data, and splicing average temperature data of a plurality of acquisition moments of the positions corresponding to the abnormal temperature data to generate time sequence temperature characteristics.
The infrared cameras regularly and continuously acquire the infrared image data of the equipment, and long-time continuous acquisition can naturally acquire the long-time sequence infrared image data of the equipment. When the equipment works normally, temperature data displayed through the infrared image data are normal, when the working state is abnormal, the temperature data corresponding to the surface of the equipment are possibly abnormal, at the moment, the abnormal temperature data at a certain acquisition moment cannot be simply analyzed, and the average temperature data in the long-time sequence infrared image data are combined for analysis, so that the performance change trend of the equipment can be analyzed.
When the infrared image data is abnormal, the corresponding position can be identified, the position is not a point, and is actually an area, so that the historical infrared image data corresponding to the position is traced back, the average temperature data corresponding to the historical infrared image data is calculated, the number of times of infrared image data is acquired, the number of times of infrared image data is corresponding, and the last average temperature data is abnormal.
The time sequence temperature characteristics are single in position, namely when average temperature data of a certain position at a certain moment is abnormal, the time sequence temperature characteristics corresponding to the position are formed by combining the historical infrared image data in a spliced mode, and each device can correspond to a plurality of time sequence temperature characteristics.
The invention relates to a method for analyzing time sequence temperature characteristics by combining a central control module with a curve fitting method, which comprises the following steps: fitting a plurality of average temperature data in the time sequence temperature characteristic based on a curve fitting method to obtain a temperature characteristic curve; identifying discrete features and trend features of the time sequence temperature features through a temperature feature curve; wherein the discrete feature represents a discrete state of the time-series temperature feature, and the trend feature represents a trend of change of the time-series temperature feature.
And fitting and acquiring a temperature characteristic curve according to the time sequence temperature characteristic, and acquiring discrete characteristics and trend characteristics of the time sequence temperature characteristic according to the temperature characteristic curve. The discrete features can be expressed by mean square error, variance and the like in mathematics, the trend features are used for judging the temperature change trend of the subsequent corresponding position according to the change of the temperature characteristic curve, and the two features can be combined to judge whether the position has risks.
In an alternative embodiment, calculating the temperature wind direction score corresponding to the time series temperature characteristic through the temperature characteristic curve includes: setting a temperature characteristic threshold value, and marking a rectangular area surrounded by the temperature characteristic threshold value in a coordinate axis as a standard area; counting the proportion of the temperature characteristic curve in the standard region, and marking the proportion as a discrete characteristic; calculating the ratio of the mean value of the first derivatives of the three acquisition moments at the end of the temperature characteristic curve to the mean value of the first derivatives of all the acquisition moments, and marking the ratio as a trend characteristic.
Setting a temperature characteristic threshold value, which is also the basis for judging whether the temperature data is abnormal, for example [40 ℃ and 60 ℃), wherein the temperature characteristic threshold value can define a rectangular area on a coordinate axis (the vertical axis is the temperature) in combination with the acquisition time, the temperature in the rectangular area accords with the working scene of the equipment, and once the temperature exceeds the rectangular area, the temperature data is abnormal. Examples statistics the proportion of the portion of the temperature profile in the standard region (i.e. the rectangular region) to the entire temperature profile, i.e. the discrete features. It can be seen that the larger the discrete feature is, the more normal the time sequence feature data is.
The first derivative of the curve can sensitively represent the curve trend, so that the ratio of the mean value of the first derivative of the temperature characteristic curve at a plurality of acquisition moments to the mean value of the first derivative of the temperature characteristic curve at all acquisition moments is used as a trend characteristic. The number of the last acquisition time is determined according to the number of samples, when the number of the samples is large, the mean value of the first derivative of the last three acquisition times can be used, and when the number of the samples is small, the value of the first derivative of the last acquisition time can be used.
The invention calculates a temperature risk score from the discrete features and the trend features, comprising: marking the discrete features as LST and the trend features as QST; obtaining a temperature risk score WFP by the formula wfp=α×|qst-qsy|/LST; where α is a scaling factor greater than 0 and QSY is a trend feature threshold.
And comprehensively considering the discrete features and the trend features to judge whether the corresponding position is abnormal or not. In the above equation obtained through extensive data simulation, the temperature risk score is proportional to the difference between the trend feature and the trend feature area, and inversely proportional to the discrete feature. The following are illustrated: setting the scaling factor α=100, qst-qsy=10, lst=20, the temperature risk score wfp=50 points for that location.
When the temperature risk score is larger than a temperature risk threshold (set according to historical experience), marking the position corresponding to the temperature risk score as an abnormal position; the central control module collects equipment appearance images through the high-definition camera, identifies material strain and fine deformation at abnormal positions based on the equipment appearance images, and calculates equipment appearance scores according to the material strain and the fine deformation.
By combining the image recognition technology with the high definition image data, the device material characteristics (such as color, degree of corrosion, etc.) and deformation characteristics (fine deformation caused by various factors) can be extracted. The invention obtains equipment appearance scores according to material strain and fine deformation, comprising the following steps: identifying whether material strain or fine deformation occurs at the abnormal position through an image identification technology; if yes, marking the risk position of the abnormal position; if not, not processing; counting the duty ratio of the corresponding areas of the plurality of risk positions on the appearance of the equipment, and obtaining the appearance score of the equipment according to the occupied ratio; wherein, the larger the risk location area, the smaller the equipment appearance score.
When a location is subject to material strain or subtle deformation, the location is marked as a risk location. And counting the proportion of the areas of the risk positions on the equipment, and obtaining the appearance score of the equipment according to the proportion. Early warning is carried out according to the appearance scores of the equipment, so that workers can process the equipment in time.
The partial data in the formula are all obtained by removing dimension and taking the numerical value for calculation, and the formula is a formula closest to the real situation obtained by simulating a large amount of collected data through software; the preset parameters and the preset threshold values in the formula are set by those skilled in the art according to actual conditions or are obtained through mass data simulation.
The working principle of the invention is as follows:
the data acquisition module acquires infrared image data with long time sequence through an infrared camera connected with the data acquisition module, and the infrared image data is preprocessed and then sent to the central control module.
The central control module extracts time sequence temperature characteristics from the infrared image data with long time sequence, analyzes the time sequence temperature characteristics by combining a curve fitting method, and obtains a temperature risk score according to an analysis result.
The central control module collects equipment appearance images based on the equipment risk scores, analyzes the equipment appearance through an image recognition technology after correcting the equipment appearance images, and obtains the equipment appearance scores according to analysis results.
The above embodiments are only for illustrating the technical method of the present invention and not for limiting the same, and it should be understood by those skilled in the art that the technical method of the present invention may be modified or substituted without departing from the spirit and scope of the technical method of the present invention.