Organic electroluminescence driving circuit fault diagnosis method based on deep learning

Abstract

The invention relates to the technical field of organic electroluminescence systems, in particular to an organic electroluminescence driving circuit fault diagnosis method based on deep learning. Comprising the following steps: establishing a plurality of display subareas according to equipment parameters of the organic light-emitting panel, establishing a plurality of monitoring points according to driving circuit parameters, and generating a display subarea-monitoring point mapping table; generating an operation deviation value of each display sub-region according to a preset monitoring time node, and selecting a plurality of to-be-diagnosed monitoring points according to all the operation deviation values; the feedback data packets of the to-be-diagnosed monitoring points are obtained, the fault diagnosis result is generated according to all the feedback data packets, the driving circuit with the fault risk is early warned in time by constructing the double diagnosis model, the driving circuit with the fault risk is rapidly positioned, the maintenance efficiency is improved, and the maintenance cost is reduced.

Description

Technical Field

[0001] The present application relates to the technical field of organic electroluminescent systems, and in particular to a fault diagnosis method for an organic electroluminescent drive circuit based on deep learning. Background Art

[0002] With the rapid development of optoelectronic technology, organic electroluminescent elements have achieved remarkable application results in the display field due to their advantages such as self-luminescence, no viewing angle limitation, fast response, power saving, simple process and full color.

[0003] An organic electroluminescent system generally includes an organic light-emitting diode array (i.e., a pixel array) and a driving circuit (i.e., a pixel circuit) that provides driving current to each organic light-emitting diode in the array. When some driving circuits operate abnormally, the operation of the organic electroluminescent system will be greatly affected. However, the current organic electroluminescent system is unable to promptly warn and accurately locate the failure risks of each driving circuit, resulting in a waste of maintenance costs for the organic electroluminescent system. Summary of the invention

[0004] The purpose of this application is: to solve the above-mentioned technical problems, this application provides an organic electroluminescent driving circuit fault diagnosis method based on deep learning, aiming to improve the early warning efficiency and diagnostic accuracy of organic electroluminescent driving circuit faults, quickly locate the driving circuit with fault risks, improve maintenance efficiency, and reduce maintenance costs.

[0005] In some embodiments of the present application, a mapping relationship between an organic light-emitting panel and each driving circuit is established, and a primary screening of the operating status of each driving circuit is performed based on the operating deviation of the organic light-emitting panel, thereby determining the driving circuit that may have a risk of failure, thereby improving the monitoring efficiency and fault diagnosis accuracy of all driving circuits.

[0006] In some embodiments of the present application, historical operating data is analyzed based on deep learning technology to extract fault characteristic parameters, thereby generating multiple monitoring indicators. According to the collected feedback data packets of each monitoring point to be diagnosed, the real-time operating data of each monitoring indicator is analyzed to generate the real-time operating status of each monitoring point to be diagnosed. By constructing a dual diagnosis model, timely warnings are issued for drive circuits with fault risks, and drive circuits with fault risks are quickly located, thereby improving maintenance efficiency and reducing maintenance costs.

[0007] In some embodiments of the present application, a method for diagnosing organic electroluminescent driving circuit faults based on deep learning is provided, comprising: Establishing multiple display sub-areas according to the device parameters of the organic light-emitting panel, establishing multiple monitoring points according to the drive circuit parameters, and generating a display sub-area-monitoring point mapping table; Generate the operation deviation value of each display sub-area according to the preset monitoring time node, and select multiple monitoring points to be diagnosed according to all the operation deviation values; Obtain feedback data packets from each monitoring point to be diagnosed, and generate fault diagnosis results based on all feedback data packets; Wherein, when establishing multiple display sub-areas, it includes: Create a display sub-area sequence A, A=(a1, a2…a i …a n ), where a i is the i-th display sub-area; n is the number of display sub-areas.

[0008] In some embodiments of the present application, when generating a display sub-area-monitoring point mapping table, the process includes: Establish a monitoring point sequence B, B = (b1, b2…b i …b m ), where b i is the i-th monitoring point; m is the number of monitoring points; According to the display sub-region sequence A, the i-th display sub-region is sequentially set as the target display sub-region; Traverse the monitoring point sequence B to generate all the associated monitoring points of the target display sub-area; Generate a mapping sub-table of the target display sub-area according to all associated monitoring points; Generate mapping sub-tables for each display sub-area in sequence; A sub-area-monitoring point mapping table is constructed based on all mapping sub-tables.

[0009] In some embodiments of the present application, when generating the running deviation value of each display sub-area, it includes: According to the display sub-region sequence A, the i-th display sub-region is sequentially set as the target display sub-region; Obtain the operating parameters of the target display sub-area at the current monitoring time node; Preprocessing of operating parameters; Generate the running deviation value d of the target display sub-area according to the preprocessing result; Generate the running deviation value of each display sub-area at the current monitoring time node in sequence; Establish the running deviation value series D of the current monitoring time node, D=(d1, d2…d i …d n ), where d i It is the running deviation value of the i-th display sub-area at the current monitoring time node.

[0010] In some embodiments of the present application, when generating the running deviation value d of the target display sub-area, it includes: d = e1 * Q1 * β 1i *(c 1i - c' 1i ) 2 + e2 * Q2 * β 2i *(c 2i - c' 2i ) 2 ; Wherein, e1 is a preset first weight coefficient; e2 is a preset second weight coefficient; Q1 is a preset first fixed coefficient; Q2 is a preset second fixed coefficient; is the number of display evaluation indicators; β 1i is the influence factor of the i-th display evaluation indicator; c 1i is the real-time reference value of the i-th display evaluation indicator in the target display sub-region at the current monitoring time node; c' 1i is the standard reference value of the i-th display evaluation indicator in the target display sub-region at the current monitoring time node; is the number of operation evaluation indicators; β 2i is the influence factor of the i-th operation evaluation indicator; c 2i is the real-time reference value of the i-th operation evaluation indicator in the target display sub-region at the current monitoring time node; c' 2i is the standard reference value of the i-th operation evaluation indicator in the target display sub-region at the current monitoring time node.

[0011] In some embodiments of the present application, when selecting multiple monitoring points to be diagnosed according to all operation deviation values, it includes: Generating a risk evaluation value k according to the operation deviation value sequence D at the current monitoring time node; Presetting a first risk evaluation value threshold K1; When k > K1, setting all monitoring points as monitoring points to be diagnosed; When k < K1, presetting a first operation deviation value threshold D1; If d i > D1, setting the associated monitoring points of the i-th display sub-region as monitoring points to be diagnosed; Obtaining all monitoring points to be diagnosed; Establishing a sequence B1 of monitoring points to be diagnosed at the current monitoring time node, B1 = (b 11 , b 12 … b 1i … b 1m1 ), wherein, b 1i is the i-th monitoring point to be diagnosed at the current monitoring time node; m1 is the number of monitoring points to be diagnosed at the current monitoring time node, and m1 ≤ m.

[0012] In some embodiments of the present application, when generating the risk assessment value k, it includes: k=e3*Q3* η i *d i ]+e4*Q4* Y(i)*η i *(d i -D1)]; Wherein, e3 is the preset third weight coefficient; e4 is the preset fourth weight coefficient; Q3 is the preset third fixed coefficient; Q4 is the preset fourth fixed coefficient; η i is the influencing factor of the i-th display sub-area; Y(i) is the selection coefficient; if (d i -D1)>0,Y(i)=1; if (d i -D1)<0,Y(i)=0.

[0013] In some embodiments of the present application, obtaining feedback data packets of each monitoring point to be diagnosed includes: Generate multiple monitoring indicators based on the preset fault diagnosis model; According to the number of monitoring points to be diagnosed, B1 is set in sequence 1i Target monitoring point to be diagnosed Generate monitoring evaluation values ​​of target monitoring points to be diagnosed according to the display sub-area-monitoring point mapping table; According to the monitoring evaluation value, the data collection amount of the target monitoring point to be diagnosed at the current monitoring time node is set; Obtain feedback data packets from target monitoring points to be diagnosed according to the amount of data collected; Obtain feedback data packets from each monitoring point to be diagnosed in turn.

[0014] In some embodiments of the present application, when generating a fault diagnosis result according to all feedback data packets, it includes: According to the feedback data packet of the target monitoring point to be diagnosed; Establish an abnormal evaluation value sequence G, G = (g1, g2…g i …g t ), where t is the number of monitoring indicators; g i is the abnormal evaluation value of the i-th monitoring indicator of the target monitoring point to be diagnosed at the current monitoring time node; Generate the fault risk value h of the target monitoring point to be diagnosed, h= w i *g i ; Among them, w i is the influencing factor of the i-th monitoring indicator; t is the number of monitoring indicators; Generate the fault risk value of each monitoring point to be diagnosed in sequence; Generate a sequence of fault risk values H at the current monitoring time node, H = (h1, h2... h i … h m1 ), where h i is the fault risk value of the i-th monitoring point to be diagnosed at the current monitoring time node; Generate a fault diagnosis result based on the sequence of fault risk values H.

[0015] In some embodiments of the present application, when establishing the sequence of abnormal evaluation values G, it includes: g i = e5 * Q5 * (f ir - f' i ) 2 + e6 * Q6 * (f ir - Δf i ) 2 ; Among them, g i is the abnormal evaluation value of the i-th monitoring index of the target monitoring point to be diagnosed at the current monitoring time node; e5 is a preset fifth weight coefficient; e6 is a preset sixth weight coefficient; Q5 is a preset fifth fixed coefficient; Q6 is a preset sixth fixed coefficient; f ir is the r-th real-time operation value collected for the i-th monitoring index at the target monitoring point to be diagnosed; f' i is the standard operation value of the i-th monitoring index at the target monitoring point to be diagnosed; Δf i is the average value of all real-time operation values collected for the i-th monitoring index at the target monitoring point to be diagnosed; s is the data acquisition volume of the target monitoring point to be diagnosed at the current monitoring time node.

[0016] In some embodiments of the present application, when generating a fault diagnosis result based on the sequence of fault risk values H, it includes: Preset a first fault risk value threshold H1 and a second fault risk value threshold H2; If h i < H1, set the i-th monitoring point to be diagnosed as a normal monitoring point; If H1 ≤ h i < H2, set the i-th monitoring point to be diagnosed as a first-level abnormal point; If h i ≥ H2, set the i-th monitoring point to be diagnosed as a second-level abnormal point; Generate a fault diagnosis result at the current monitoring time node based on all first-level abnormal points and all second-level abnormal points.

[0017] Compared with the prior art, the beneficial effect of a fault diagnosis method for an organic electroluminescence driving circuit based on deep learning in an embodiment of the present application is: By establishing a mapping relationship between the organic light-emitting panel and each driving circuit, the operating status of each driving circuit is initially screened based on the operating deviation of the organic light-emitting panel, thereby determining the driving circuit that may have a risk of failure, thereby improving the monitoring efficiency and fault diagnosis accuracy of all driving circuits.

[0018] Based on deep learning technology, historical operation data is analyzed to extract fault characteristic parameters, thereby generating multiple monitoring indicators. According to the collected feedback data packets of each monitoring point to be diagnosed, the real-time operation data of each monitoring indicator is analyzed to generate the real-time operation status of each monitoring point to be diagnosed. By building a dual diagnosis model, timely warnings are given to drive circuits with fault risks, and drive circuits with fault risks are quickly located, thereby improving maintenance efficiency and reducing maintenance costs. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] Figure 1 It is a flow chart of a method for fault diagnosis of an organic electroluminescent driving circuit based on deep learning in a preferred embodiment of the present application. DETAILED DESCRIPTION

[0020] The specific implementation methods of the present application are further described in detail below in conjunction with the accompanying drawings and examples. The following examples are used to illustrate the present application but are not intended to limit the scope of the present application.

[0021] In the description of the present application, it should be understood that the terms "center", "up", "down", "front", "back", "left", "right", "vertical", "horizontal", "top", "bottom", "inside", "outside", etc., indicating orientations or positional relationships, are based on the orientations or positional relationships shown in the accompanying drawings, and are only for the convenience of describing the present application and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore should not be understood as a limitation on the present application.

[0022] The terms "first" and "second" are used for descriptive purposes only and should not be understood as indicating or implying relative importance or implicitly indicating the number of the indicated technical features. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of the features. In the description of this application, unless otherwise specified, "plurality" means two or more.

[0023] In the description of this application, it should be noted that, unless otherwise clearly specified and limited, the terms "installed", "connected", and "connected" should be understood in a broad sense, for example, it can be a fixed connection, a detachable connection, or an integral connection; it can be a mechanical connection or an electrical connection; it can be a direct connection, or it can be indirectly connected through an intermediate medium, or it can be the internal communication of two components. For ordinary technicians in this field, the specific meanings of the above terms in this application can be understood according to specific circumstances.

[0024] like Figure 1 As shown, a method for diagnosing an organic electroluminescent driving circuit fault based on deep learning in a preferred embodiment of the present application includes: S101: establishing a plurality of display sub-areas according to device parameters of the organic light-emitting panel, establishing a plurality of monitoring points according to drive circuit parameters, and generating a display sub-area-monitoring point mapping table; S102: generating an operation deviation value of each display sub-area according to a preset monitoring time node, and selecting a plurality of monitoring points to be diagnosed according to all the operation deviation values; S103: Obtaining feedback data packets of each monitoring point to be diagnosed, and generating a fault diagnosis result according to all feedback data packets; Wherein, when establishing multiple display sub-areas, it includes: Create a display sub-area sequence A, A=(a1, a2…a i …a n ), where a i is the i-th display sub-area; n is the number of display sub-areas.

[0025] Specifically, the number of display sub-areas can be set according to the number of driving circuits in the current organic electroluminescent system and the area of ​​the organic light-emitting panel.

[0026] Specifically, the feature is that when generating and displaying the sub-area-monitoring point mapping table, it includes: Establish a monitoring point sequence B, B = (b1, b2…b i …b m ), where b i is the i-th monitoring point; m is the number of monitoring points; According to the display sub-region sequence A, the i-th display sub-region is sequentially set as the target display sub-region; Traverse the monitoring point sequence B to generate all the associated monitoring points of the target display sub-area; Generate a mapping sub-table of the target display sub-area according to all associated monitoring points; Generate mapping sub-tables for each display sub-area in sequence; A sub-area-monitoring point mapping table is constructed based on all mapping sub-tables.

[0027] Specifically, a single monitoring point represents a driving circuit. If there is a connection relationship between the driving circuit of the current monitoring point and the target display sub-area, the monitoring point is set as a management monitoring point, and a display sub-area-monitoring point mapping table is constructed based on the connection relationship between the driving circuit and each display sub-area, wherein there is a many-to-many mapping relationship between the monitoring point and the display sub-area.

[0028] In a preferred embodiment of the present application, when generating the running deviation value of each display sub-area, it includes: According to the display sub-region sequence A, the i-th display sub-region is sequentially set as the target display sub-region; Obtain the operating parameters of the target display sub-area at the current monitoring time node; Preprocessing of operating parameters; Generate the running deviation value d of the target display sub-area according to the preprocessing result; Generate the running deviation value of each display sub-area at the current monitoring time node in sequence; Establish the running deviation value series D of the current monitoring time node, D=(d1, d2…d i …d n ), where d i It is the running deviation value of the i-th display sub-area at the current monitoring time node.

[0029] Specifically, when generating the running deviation value d of the target display sub-area, it includes: d=e1*Q1*[ β 1i *(c 1i -c' 1i ) 2 ]+e2*Q2*[ β 2i *(c 2i -c' 2i ) 2 ]; Wherein, e1 is the preset first weight coefficient; e2 is the preset second weight coefficient; Q1 is the preset first fixed coefficient; Q2 is the preset second fixed coefficient; To display the number of evaluation indicators; β 1i is the influencing factor of the i-th display evaluation index; c 1i is the real-time reference value of the i-th display evaluation index in the target display sub-area at the current monitoring time node; c' 1i is the standard reference value of the i-th display evaluation index in the target display sub-area at the current monitoring time node; is the number of operation evaluation indicators; β 2i is the influencing factor of the i-th operation evaluation index; c 2iis the real-time reference value of the i-th operation evaluation index in the target display sub-region at the current monitoring time node; c' 2i is the standard reference value of the i-th operation evaluation index in the target display sub-region at the current monitoring time node.

[0030] Specifically, all parameters in the model are normalized by presetting a first fixed coefficient and a second fixed coefficient, so that each parameter is within the same value range.

[0031] Specifically, the display evaluation indexes include but are not limited to parameters such as peak brightness, average brightness, color ratio, etc., and the operation evaluation indexes include but are not limited to current value, temperature value of the display sub-region, whether there is a color deviation trajectory, ghosting, screen flashing, etc. The real-time reference values of each display evaluation index are generated by analyzing the real-time display data of the target display sub-region, and the standard reference values of each display evaluation index in the target display sub-region are generated according to the user display requirements at the current monitoring time node.

[0032] Specifically, the larger the operation deviation value is, the more unstable the operation state of the current display sub-region is.

[0033] Specifically, each influencing factor can be set according to historical parameters. The larger the influencing factor is, the greater the impact on the stable operation of the target display sub-region when the corresponding index fluctuates.

[0034] Specifically, when selecting multiple monitoring points to be diagnosed according to all operation deviation values, it includes: Generating a risk evaluation value k according to the operation deviation value sequence D at the current monitoring time node; Presetting a first risk evaluation value threshold K1; When k > K1, setting all monitoring points as monitoring points to be diagnosed; When k < K1, presetting a first operation deviation value threshold D1; If d i > D1, setting the associated monitoring point of the i-th display sub-region as a monitoring point to be diagnosed; Obtaining all monitoring points to be diagnosed; Establishing a sequence B1 of monitoring points to be diagnosed at the current monitoring time node, B1 = (b 11 , b 12 …b 1i …b 1m1 ), where b 1i is the i-th monitoring point to be diagnosed at the current monitoring time node; m1 is the number of monitoring points to be diagnosed at the current monitoring time node, and m1 ≤ m.

[0035] Specifically, the first risk assessment value threshold K1 and the first operation deviation value threshold D1 are both set according to historical parameters. When k>K1, it means that the current organic electroluminescent system is in abnormal operation and fault diagnosis needs to be performed on all monitoring points. i >D1, it means that the i-th display sub-area is in an abnormal operating state, and the corresponding driving circuit is likely to have a high risk of failure.

[0036] Specifically, by monitoring the operating status of each display sub-area in real time, it is possible to promptly determine whether each monitoring point has a risk of abnormal operation, thereby quickly determining the corresponding monitoring point to be diagnosed, thereby improving the fault diagnosis efficiency of the drive circuit.

[0037] Specifically, when generating the risk assessment value k, it includes: k=e3*Q3* η i *d i ]+e4*Q4* Y(i)*η i *(d i -D1)]; Wherein, e3 is the preset third weight coefficient; e4 is the preset fourth weight coefficient; Q3 is the preset third fixed coefficient; Q4 is the preset fourth fixed coefficient; η i is the influencing factor of the i-th display sub-area; Y(i) is the selection coefficient; if (d i -D1)>0,Y(i)=1; if (d i -D1)<0,Y(i)=0.

[0038] Specifically, by presetting the third fixed coefficient and the fourth fixed coefficient, all parameters in the model are normalized so that each parameter is within the same value range.

[0039] It can be understood that in the above embodiment, by establishing a mapping relationship between the organic light-emitting panel and each driving circuit, the operating status of each driving circuit is initially screened based on the operating deviation of the organic light-emitting panel, thereby determining the driving circuit that may have a risk of failure, thereby improving the monitoring efficiency and fault diagnosis accuracy of all driving circuits.

[0040] In a preferred embodiment of the present application, when obtaining feedback data packets of each monitoring point to be diagnosed, it includes: Generate multiple monitoring indicators based on the preset fault diagnosis model; According to the number of monitoring points to be diagnosed, B1 is set in sequence 1i Target monitoring point to be diagnosed Generate monitoring evaluation values ​​of target monitoring points to be diagnosed according to the display sub-area-monitoring point mapping table; According to the monitoring evaluation value, the data collection amount of the target monitoring point to be diagnosed at the current monitoring time node is set; Obtain feedback data packets from target monitoring points to be diagnosed according to the amount of data collected; Obtain feedback data packets from each monitoring point to be diagnosed in turn.

[0041] Specifically, its monitoring evaluation value can be set according to the sum of the operating deviation values ​​of all mapped display sub-areas of the target monitoring point to be diagnosed. The larger the sum, the larger the corresponding monitoring evaluation value. The larger the monitoring evaluation value, the greater the possibility of failure risk of the target monitoring point to be diagnosed, and the larger the corresponding data collection amount.

[0042] Specifically, the monitoring indicators include but are not limited to temperature, current value, voltage value and other fault characteristic parameters extracted by analyzing historical operation data based on deep learning technology.

[0043] Specifically, by dynamically adjusting the amount of data collected at each monitoring point to be diagnosed, the interference of accidental data on the diagnosis results is eliminated, thereby ensuring the accuracy of fault diagnosis for each monitoring point.

[0044] Specifically, when generating fault diagnosis results based on all feedback data packets, it includes: According to the feedback data packet of the target monitoring point to be diagnosed; Establish an abnormal evaluation value sequence G, G = (g1, g2…g i …g t ), where t is the number of monitoring indicators; g i is the abnormal evaluation value of the i-th monitoring indicator of the target monitoring point to be diagnosed at the current monitoring time node; Generate the fault risk value h of the target monitoring point to be diagnosed, h= w i *g i ; Among them, w i is the influencing factor of the i-th monitoring indicator; Generate the fault risk value of each monitoring point to be diagnosed in turn; t is the number of monitoring indicators; Generate the fault risk value sequence H of the current monitoring time node, H=(h1,h2…h i …h m1 ), where h i is the fault risk value of the i-th monitoring point to be diagnosed at the current monitoring time node; Generate fault diagnosis results based on the fault risk value sequence H.

[0045] Specifically, the greater the fault risk value, the greater the possibility that the drive circuit of the corresponding monitoring point has a fault.

[0046] Specifically, when establishing the abnormal evaluation value sequence G, it includes: g i = e5 * Q5 * (f ir - f' i ) 2 + e6 * Q6 * (f ir - Δf i ) 2 ; Among them, g i is the abnormal evaluation value of the i-th monitoring index of the target monitoring point to be diagnosed at the current monitoring time node; e5 is the preset fifth weight coefficient; e6 is the preset sixth weight coefficient; Q5 is the preset fifth fixed coefficient; Q6 is the preset sixth fixed coefficient; f ir is the r-th real-time operation value collected for the i-th monitoring index at the target monitoring point to be diagnosed; f' i is the standard operation value of the i-th monitoring index at the target monitoring point to be diagnosed; Δf i is the average value of all real-time operation values collected for the i-th monitoring index at the target monitoring point to be diagnosed; s is the data acquisition volume of the target monitoring point to be diagnosed at the current monitoring time node.

[0047] Specifically, all parameters in the model are normalized by the preset fifth fixed coefficient and sixth fixed coefficient, so that each parameter is in the same value range.

[0048] Specifically, the standard operation value of each monitoring index at the target monitoring point to be diagnosed can be set according to historical parameters.

[0049] Specifically, based on deep learning technology, historical operation data is analyzed to extract fault feature parameters, thereby generating multiple monitoring indexes. According to the feedback data packets collected from each monitoring point to be diagnosed, the real-time operation data of each monitoring index is analyzed to generate the real-time operation status of each monitoring point to be diagnosed. By constructing a dual diagnosis model, early warnings are given to the drive circuits at risk of failure in a timely manner, the drive circuits at risk of failure are quickly located, the maintenance efficiency is improved, and the maintenance cost is reduced.

[0050] Specifically, when generating the fault diagnosis result according to the fault risk value sequence H, it includes: Presetting the first fault risk value threshold H1 and the second fault risk value threshold H2; If h i < H1, set the i-th monitoring point to be diagnosed as a normal monitoring point; If H1 ≤ h i < H2, set the i-th monitoring point to be diagnosed as a first-level abnormal point; If hi ≥H2, set the i-th monitoring point to be diagnosed as a secondary abnormal point; Generate the fault diagnosis result of the current monitoring time node based on all the first-level abnormal points and all the second-level abnormal points.

[0051] Specifically, a first-level abnormal point indicates that the driving circuit of the current monitoring point is operating abnormally. The corresponding maintenance plan can be set according to the current number of first-level abnormal points. A second-level abnormal point indicates that the driving circuit corresponding to the current monitoring point may have an operational failure at any time, thereby affecting the operation of the system. Timely maintenance is required to eliminate the risk of failure.

[0052] According to the first concept of the present application, a mapping relationship between the organic light-emitting panel and each driving circuit is established, and the operating status of each driving circuit is initially screened based on the operating deviation of the organic light-emitting panel, so as to determine the driving circuit that may have a risk of failure, thereby improving the monitoring efficiency and fault diagnosis accuracy of all driving circuits.

[0053] According to the second concept of the present application, historical operating data is analyzed based on deep learning technology to extract fault characteristic parameters, thereby generating multiple monitoring indicators. According to the collected feedback data packets of each monitoring point to be diagnosed, the real-time operating data of each monitoring indicator is analyzed to generate the real-time operating status of each monitoring point to be diagnosed. By constructing a dual diagnosis model, timely warnings can be given to drive circuits with fault risks, and drive circuits with fault risks can be quickly located, thereby improving maintenance efficiency and reducing maintenance costs.

[0054] The above is only a preferred implementation of the present application. It should be pointed out that for ordinary technicians in this technical field, several improvements and substitutions can be made without departing from the technical principles of the present application. These improvements and substitutions should also be regarded as the scope of protection of the present application.

Claims

1. A method for fault diagnosis of an organic electroluminescent drive circuit based on deep learning, characterized in that: Including: Establish multiple display sub-regions according to the device parameters of the organic light-emitting panel, establish multiple monitoring points according to the drive circuit parameters, and generate a display sub-region - monitoring point mapping table; Generate the operation deviation value of each display sub-region according to the preset monitoring time node, and select multiple monitoring points to be diagnosed according to all the operation deviation values; Obtain the feedback data packets of each monitoring point to be diagnosed, and generate a fault diagnosis result according to all the feedback data packets; Among them, when establishing multiple display sub-regions, it includes: Create a display sub-area sequence A, A=(a1, a2…a i …a n ), where a i is the i-th display sub-area; n is the number of display sub-areas.

2. The organic electroluminescent drive circuit fault diagnosis method based on deep learning as claimed in claim 1, characterized in that: When generating the display sub-region - monitoring point mapping table, it includes: Establish a monitoring point sequence B, B = (b1, b2…b i …b m ), where b i is the i-th monitoring point; m is the number of monitoring points; Set the i-th display sub-region as the target display sub-region in sequence according to the display sub-region sequence A; Traverse the monitoring point sequence B to generate all the associated monitoring points of the target display sub-region; Generate a mapping sub-table of the target display sub-region according to all the associated monitoring points; Generate the mapping sub-tables of each display sub-region in sequence; Construct a display sub-region - monitoring point mapping table according to all the mapping sub-tables.

3. The organic electroluminescent drive circuit fault diagnosis method based on deep learning as claimed in claim 2, characterized in that: When generating the operation deviation value of each display sub-region, it includes: Set the i-th display sub-region as the target display sub-region in sequence according to the display sub-region sequence A; Obtain the operation parameters of the target display sub-region at the current monitoring time node; Preprocess the operation parameters; Generate the operation deviation value d of the target display sub-region according to the preprocessing result; Generate the operation deviation values of each display sub-region at the current monitoring time node in sequence; Establish the running deviation value series D of the current monitoring time node, D=(d1, d2…d i …d n ), where d i It is the running deviation value of the i-th display sub-area at the current monitoring time node.

4. The organic electroluminescent drive circuit fault diagnosis method based on deep learning as claimed in claim 3, characterized in that: When generating the operation deviation value d of the target display sub-region, it includes: d=e1*Q1*[ β 1i *(c 1i -c' 1i ) ) 2 ]+e2*Q2*[ β 2i *(c 2i -c' 2i ) ) 2 ]; Among them, e1 is the preset first weight coefficient; e2 is the preset second weight coefficient; Q1 is the preset first fixed coefficient; Q2 is the preset second fixed coefficient; To display the number of evaluation indicators; β 1i is the influencing factor of the i-th display evaluation index; c 1i is the real-time reference value of the i-th display evaluation index in the target display sub-area at the current monitoring time node; c' 1i is the standard reference value of the i-th display evaluation index in the target display sub-area at the current monitoring time node; is the number of operation evaluation indicators; β 2i is the influencing factor of the i-th operation evaluation index; c 2i c' is the real-time reference value of the i-th operation evaluation index in the target display sub-area at the current monitoring time node; 2i The standard reference value of the i-th operation evaluation index in the target display sub-area at the current monitoring time node.

5. The organic electroluminescent drive circuit fault diagnosis method based on deep learning as claimed in claim 3, characterized in that: When selecting multiple monitoring points to be diagnosed according to all the operation deviation values, it includes: Generate a risk evaluation value k according to the operation deviation value sequence D at the current monitoring time node; Preset the first risk evaluation value threshold K1; When k > K1, set all the monitoring points as the monitoring points to be diagnosed; When k < K1, preset the first operation deviation value threshold D1; If d i >D1, set the associated monitoring point of the i-th display sub-area as the monitoring point to be diagnosed; Obtain all the monitoring points to be diagnosed; Establish the number of monitoring points to be diagnosed at the current monitoring time node B1, B1=(b 11 , b 12 …b 1i …b 1m1 ), where b 1i is the i-th monitoring point to be diagnosed at the current monitoring time node; m1 is the number of monitoring points to be diagnosed at the current monitoring time node, and m1≤m.

6. The organic electroluminescent drive circuit fault diagnosis method based on deep learning as claimed in claim 5, characterized in that: When generating the risk evaluation value k, it includes: k=e3*Q3* η i *d i ]+e4*Q4* Y(i)*η i *(d i -D1)]; Wherein, e3 is the preset third weight coefficient; e4 is the preset fourth weight coefficient; Q3 is the preset third fixed coefficient; Q4 is the preset fourth fixed coefficient; η i is the influencing factor of the i-th display sub-area; Y(i) is the selection coefficient; if (d i -D1)>0,Y(i)=1; if (d i -D1)<0,Y(i)=0.

7. The organic electroluminescent drive circuit fault diagnosis method based on deep learning as claimed in claim 5, characterized in that: When obtaining the feedback data packets of each monitoring point to be diagnosed, it includes: Generate multiple monitoring indicators according to the preset fault diagnosis model; According to the number of monitoring points to be diagnosed, B1 is set in sequence 1i Target monitoring point to be diagnosed Generate the monitoring evaluation value of the target monitoring point to be diagnosed according to the display sub-region - monitoring point mapping table; Set the data acquisition amount of the target monitoring point to be diagnosed at the current monitoring time node according to the monitoring evaluation value; Obtain the feedback data packet of the target monitoring point to be diagnosed according to the data acquisition amount; Obtain the feedback data packets of each monitoring point to be diagnosed in sequence.

8. The organic electroluminescent drive circuit fault diagnosis method based on deep learning as claimed in claim 7, characterized in that: When generating a fault diagnosis result according to all the feedback data packets, it includes: According to the feedback data packet of the target monitoring point to be diagnosed; Establish an abnormal evaluation value sequence G, G = (g1, g2…g i …g t ), where t is the number of monitoring indicators; g i is the abnormal evaluation value of the i-th monitoring indicator of the target monitoring point to be diagnosed at the current monitoring time node; Generate the fault risk value h of the target monitoring point to be diagnosed, h= w i *g i ; Among them, w i is the influencing factor of the i-th monitoring indicator; t is the number of monitoring indicators; Generate the fault risk values of each monitoring point to be diagnosed in sequence; Generate the fault risk value sequence H of the current monitoring time node, H=(h1,h2…h i …h m1 ), where h i is the fault risk value of the i-th monitoring point to be diagnosed at the current monitoring time node; Generate a fault diagnosis result according to the fault risk value sequence H.

9. The organic electroluminescent drive circuit fault diagnosis method based on deep learning as claimed in claim 8, characterized in that: When establishing the abnormal evaluation value sequence G, it includes: g i =e5*Q5*[ (f ir -f' i ) 2 ]+e6*Q6*[ (f ir -Δf i ) 2 ]; Among them, g i is the abnormal evaluation value of the i-th monitoring indicator of the target monitoring point to be diagnosed at the current monitoring time node; e5 is the preset fifth weight coefficient; e6 is the preset sixth weight coefficient; Q5 is the preset fifth fixed coefficient; Q6 is the preset sixth fixed coefficient; f ir is the rth real-time running value collected at the target monitoring point for diagnosis for the i-th monitoring indicator; f' i is the standard operating value of the i-th monitoring indicator at the target monitoring point to be diagnosed; Δf i is the average value of all real-time operating values ​​collected by the i-th monitoring indicator at the target monitoring point to be diagnosed; s is the data collection amount of the target monitoring point to be diagnosed at the current monitoring time node.

10. The organic electroluminescent drive circuit fault diagnosis method based on deep learning according to claim 9, characterized in that: When generating a fault diagnosis result according to the fault risk value sequence H, it includes: Preset the first fault risk value threshold H1 and the second fault risk value threshold H2; If h i <H1, set the i-th monitoring point to be diagnosed as a normal monitoring point; If H1 ≤ h i <H2, set the i-th monitoring point to be diagnosed as a first-level abnormal point; If h i ≥H2, set the i-th monitoring point to be diagnosed as a secondary abnormal point; Generate the fault diagnosis result at the current monitoring time node according to all the first-level abnormal points and all the second-level abnormal points.

Images