An intelligent detection and quality control process for a special high voltage porcelain insulator

Abstract

The application discloses a full-process intelligent detection and quality control process for extra-high voltage porcelain insulators, and particularly relates to the field of electrical detection and quality control, which comprises reading the umbrella skirt number, medium section and electrical measurement position of a target extra-high voltage porcelain insulator, dividing the medium section into a slip section, a cleaning liquid section and a cleaning liquid section before gluing, generating a leakage position map according to the pointing relationship from the slip section to the glaze backflow line position, from the cleaning liquid section to the umbrella skirt root position and from the cleaning liquid section before gluing to the gluing edge position; and projecting the differential results of the medium section double-end trace detection to the glaze backflow line position, the umbrella skirt root position and the gluing edge position, and combining the three-time electrical measurement difference, non-return path search and point deletion replacement counterproof to determine the effective leakage path position, so as to solve the problems that the front-end trace residues cannot be determined whether to occupy the subsequent leakage path position, and it is difficult to determine the re-inspection position and interception range after the finished product is abnormal.

Description

Technical Field

[0001] This invention relates to the field of electrical testing and quality control technology, and more specifically, to a process for intelligent testing and quality control of ultra-high voltage porcelain insulators throughout the entire process. Background Technology

[0002] In the quality control of UHV porcelain insulator production, existing tests mainly focus on raw material purity, blank appearance, glaze defects, dimensional deviation, mechanical strength, and finished product electrical performance. The front end uses trace detection through the conductivity or ion response of slurry, glaze, and cleaning fluid, while the back end uses insulation resistance, power frequency withstand voltage, partial discharge, or dielectric loss to determine whether the finished product meets the factory requirements. In the continuous production of multi-skirt porcelain insulators, the same glaze slurry tank, cleaning circulation section, or pre-gluing cleaning fluid will repeatedly come into contact with multiple blanks. The root of the skirt, the glaze return line, the micropores of the blank opening, and the gluing boundary are all locations where leakage channels are likely to form under subsequent damp or contaminated conditions. At this time, relying solely on the liquid phase trace detection results can only determine whether there are ion residues in the contact medium, and relying solely on the finished product electrical test results can only determine whether the insulation performance is abnormal. It is impossible to distinguish whether the trace residues have only passed through the surface of the porcelain part or have entered the location where leakage paths may form later. This defect manifests in the production line as a lack of interception basis when the front-end trace detection is slightly abnormal, and it is difficult to reverse the inspection location and the scope of affected products after the finished product electrical abnormality occurs. This leads to quality control repeatedly switching between front-end release and back-end expanded re-inspection. Therefore, how to determine whether trace residue occupies an effective leakage path location based on the correlation between trace detection results and local electrical parameter changes of porcelain insulators has become the technical problem to be solved in this application. Summary of the Invention

[0003] To overcome the aforementioned deficiencies of the prior art, embodiments of the present invention provide a full-process intelligent detection and quality control process for ultra-high voltage porcelain insulators. This process projects the differential results of double-end trace detection of the dielectric section onto the glaze return line position, the umbrella skirt root position, and the glued edge position. It also combines the three-time electrical measurement difference, non-return path search, and point deletion substitution to determine the effective leakage path location, thereby solving the problems mentioned in the background art.

[0004] To achieve the above objectives, the present invention provides the following technical solution: a fully intelligent detection and quality control process for ultra-high voltage porcelain insulators, comprising: S1. Read the skirt number, dielectric section and electrical measurement position of the target UHV porcelain insulator. Divide the dielectric section into glaze slurry section, cleaning liquid section and pre-applied cleaning liquid section. Generate a leakage location map according to the pointing relationship from the glaze slurry section to the glaze surface return line position, the cleaning liquid section to the skirt root position, and the pre-applied cleaning liquid section to the appliqué edge position. S2. Perform double-end trace detection on the medium segment, calculate the ion response increment and conductivity response increment of the same medium segment, record the medium segments with the same sign and non-zero as residual points, and write the residual points into the leakage position pointed to by the same medium segment through arc length projection calculation to generate residual projection record. S3. Read the three-time electrical measurement values ​​of the leak location and the adjacent surface location in the residual projection record, calculate the contact difference and recovery difference, and record the records where the leak location has the same contact difference and the same recovery difference, and the adjacent surface location does not have the same recovery difference as the retention point. S4. Using the residual point as the starting point and the stagnation point as the ending point, expand the path along the pointing relationship of the leakage bit map according to the non-return graph search algorithm. Stop the path that enters the adjacent surface bit, returns to the leaked bit or has no subsequent leaked bit and does not reach the stagnation point, and generate candidate paths. S5. Recalculate the connection state of the candidate path by the deletion point splitting and pruning algorithm, perform residual substitution and lingering substitution proof on the splitting candidate path, record the splitting candidate path that was not substituted as a placeholder path, and output the effective leakage path location, re-inspection location and interception range.

[0005] In a preferred embodiment, S1 includes: S1-1. Read the outer edge line of the umbrella skirt, the upper surface line of the umbrella skirt, the lower surface line of the umbrella skirt, the outer edge line of the fitting, and the glue filling edge line from the axial profile data of the target UHV porcelain insulator. Record the glaze flow termination line as the glaze return line position, the intersection line of the upper surface line of the umbrella skirt and the outer contour of the porcelain core as the umbrella skirt root position, the intersection line of the glue filling edge line and the outer contour of the porcelain end as the glue edge position, and record the surface sampling line of a single adjacent sampling position from the glaze return line position, the umbrella skirt root position, or the glue edge position as the adjacent surface position, and generate an electrical measurement position table. S1-2. For the medium segment, read the medium type, the shed number, the arc length of the contact start point, and the arc length of the contact end point. The arc length of the contact start point is the curved arc length from the starting point of the contact trajectory along the same shed surface sampling line to the root position when the medium enters the surface of the target UHV porcelain insulator. The arc length of the contact end point is the curved arc length from the end point of the contact trajectory along the same shed surface sampling line to the root position when the medium leaves the surface of the target UHV porcelain insulator. The medium segment with the medium type being glaze slurry and the arc length of the contact end point falling within the arc length range of the glaze return line is recorded as the glaze slurry segment. The medium segment with the medium type being cleaning fluid and the arc lengths of the contact start point and the contact end point being located on both sides of the arc length at the root position of the shed is recorded as the cleaning fluid segment. The medium segment with the medium type being pre-applied cleaning fluid and the arc length of the contact end point falling within the arc length range of the appliqué edge is recorded as the pre-applied cleaning fluid segment. Generate a medium segmentation table. S1-3. Based on the electrical measurement position table and the medium segmentation table, connect the glaze slurry segment to the glaze return line position under the same umbrella skirt number, connect the cleaning liquid segment to the umbrella skirt root position under the same umbrella skirt number, connect the pre-applied cleaning liquid segment to the applicator edge position under the same umbrella skirt number, and connect each adjacent surface position to the adjacent glaze return line position, umbrella skirt root position, or applicator edge position under the same umbrella skirt number to generate a leak position map.

[0006] In a preferred embodiment, S2 includes: S2-1. For each dielectric segment, take an inlet sample before the dielectric enters the surface of the porcelain insulator and an outlet sample after the dielectric leaves the surface of the porcelain insulator. Encode the target ion type of the ion-selective electrode as the ion channel number, record the millivolt value output by the ion-selective electrode as the ion electrode potential value, and record the micro-Siemens per centimeter value output by the conductivity electrode as the conductivity value, and generate a segment sample record. S2-2. Based on segment sample records, a two-end difference algorithm is used to subtract the ion electrode potential value of the inlet sample from the ion electrode potential value of the outlet sample under the same ion channel number to obtain the ion channel increment, and subtract the conductance value of the inlet sample from the conductance value of the outlet sample to obtain the conductance response increment, thus generating segment difference records.

[0007] In a preferred embodiment, S2 further includes: S2-3. Using the same-sign rank selection algorithm, retain the ion channel increments with the same sign as the conductivity response increment, arrange them in descending order of the absolute value of the retained ion channel increments, write the ion channel increment at the top of the sort as the ion response increment, and record the medium segment where both the ion response increment and the conductivity response increment are not zero as the residual point. S2-4. Based on the residual point, medium type, flow through the umbrella skirt number, contact start arc length, contact end arc length, and leakage location map, the arc length projection algorithm is used. First, the leakage location under the same umbrella skirt number is selected according to the flow through the umbrella skirt number. Then, the residual point is written according to the medium type: the residual point of the glaze slurry section is written as the glaze return line position with an arc length range including the contact end arc length; the residual point of the cleaning liquid section is written as the umbrella skirt root position with an arc length range between the contact start arc length and the contact end arc length; and the residual point of the pre-gluing cleaning liquid section is written as the gluing edge position with an arc length range including the contact end arc length. This generates a residual projection record.

[0008] In a preferred embodiment, S3 includes: S3-1. Read the leakage position written by the residual point according to the residual projection record, and read the adjacent surface position connected to the leakage position from the leakage position map. Read the local capacitance picofarad value, low frequency impedance amplitude ohm value and insulation resistance megohm value of the leakage position and adjacent surface position respectively during the pre-contact test, post-contact test and recovery test, and generate the position test record. S3-2. Based on the position measurement record, a three-time difference algorithm is used to subtract the corresponding measured values ​​from the pre-contact measurement values ​​from the local capacitance picofarad value, low-frequency impedance amplitude ohm value, and insulation resistance megohm value measured after contact to obtain the contact difference value. The corresponding measured values ​​from the pre-contact measurement value are then subtracted from the corresponding measured values ​​during the recovery measurement to obtain the recovery difference value, thus generating the position difference record.

[0009] In a preferred embodiment, S3 further includes: S3-3. Using the three-symbol parallel algorithm, read the position difference record of the leakage position, and write the measurement items in the local capacitance picofarad value, low frequency impedance amplitude ohm value and insulation resistance megohm value, where the contact difference and recovery difference are all non-zero and have the same sign, as leakage retention items, and bind the leakage retention items to the leakage position to generate leakage retention records. S3-4. Using the neighboring position proof algorithm, the recovery difference sign of the leakage retention item is compared with the recovery difference sign of the corresponding measurement item of the adjacent surface position. The leakage position of the adjacent surface position that does not have the same name and the same sign recovery difference is recorded as the retention point.

[0010] In a preferred embodiment, S4 includes: S4-1. Read the leak bits written by the residual points according to the residual projection record, record the residual points as the road start, record the leak bits as the first leak bit, and generate the initial road string by the road start and the first leak bit in the writing order. S4-2. Based on the initial path string, read the pointing nodes of the leak bits at the end of the path string in the leak bit graph along the pointing relationship according to the non-backward graph search algorithm, write the leak bits that have not been written into the path string as the bits to be traversed, and write the adjacent surface bits and the branches that have been written into the path string as the branches to be stopped.

[0011] In a preferred embodiment, S4 further includes: S4-3. For each position to be moved, add each position to the end of the path string and then repeat the node pointed to by the leaked position at the end of the path string until the leaked position at the end of the path string is the same as the stagnation point, and then generate the arrival path string. S4-4. Stop extending the path when the position to be moved is empty and the leakage position at the end of the path is different from the stagnation point. Record the path that arrives with the stagnation point at the beginning of the path and the stagnation point at the end of the path as a candidate path.

[0012] In a preferred embodiment, S5 includes: S5-1. Based on the candidate path, a three-deletion splitting algorithm is adopted. The residual points, each leakage point and the stagnation point in the candidate path are respectively regarded as deletion objects. After removing one deletion object at a time, the non-return graph search is recalculated. If the residual point is disconnected from the stagnation point, the disconnection position is written. If the residual point reaches the stagnation point, the connection position is written, and the deletion point bit string is generated. S5-2. Based on the deletion point bit string, use the major bit determination algorithm. If the deletion of the leaked bit results in a disconnected bit, the deletion of the residual point results in a disconnected bit, and the deletion of the stagnant point results in a disconnected bit, then write the candidate path into the split candidate path; otherwise, write the candidate path into the non-occupied path.

[0013] In a preferred embodiment, S5 further includes: S5-3. For the candidate path of the split, the double substitution proof algorithm is adopted. The residual points other than the current residual point in the same medium segment are used as residual substitution points. The non-return graph search from the residual substitution point to the current stagnation point is recalculated and a residual substitution record is generated. The stagnation points other than the current stagnation point under the same leakage position are used as stagnation substitution points. The non-return graph search from the current residual point to the stagnation substitution point is recalculated and a stagnation substitution record is generated. S5-4. If the residual or delayed replacement record is a connected bit, the candidate path for severance is written into the replacement hit record. Otherwise, the candidate path for severance is written into the pending occupancy path. The lexicographical pruning and write-back algorithm is used to select the occupancy path in ascending order of the number of leakage bits, descending order of the number of disconnection bits, and ascending order of candidate path writing. The leakage bits in the occupancy path are written as the effective leakage path location. The umbrella skirt number and electrical measurement bit to which the effective leakage path location belongs are written as the re-inspection location. The target UHV porcelain insulator with the same medium segment as the residual point in the occupancy path is written as the interception range.

[0014] The technical effects and advantages of this invention are as follows: 1. By projecting the trace detection results onto the leakage location via arc length, the residual ions are associated with the umbrella skirt root, glaze return line, and glue edge, which relatively improves the problem of lacking interception evidence for minor anomalies at the front end. 2. By calculating the contact difference and recovery difference through three-time electrical measurement values ​​and introducing the adjacent surface position as evidence, it is possible to distinguish between leakage site retention and surface residue in the same direction, thereby relatively reducing the probability of misjudging the re-inspection location; 3. By using non-return map search to limit the one-way arrival path from the residual point to the stagnation point, the candidate path avoids adjacent surface positions and return leakage positions, reducing the number of surface detour paths entering the occupancy judgment. 4. By deleting points, pruning, and recalculating the connectivity status of candidate paths, it is possible to determine whether a leak point plays a necessary role, making the determination of the effective leak path location more structurally based. 5. By using residual substitution and retention substitution to refute the substitution explanation, and then rewriting the re-inspection location and interception range, the repeated switching between expanding re-inspection and front-end release after the finished product is abnormal can be relatively alleviated. Attached Figure Description

[0015] Figure 1 This is a flowchart of the present invention. Detailed Implementation

[0016] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0017] Refer to the instruction manual appendix Figure 1 The present invention provides a fully intelligent detection and quality control process for ultra-high voltage porcelain insulators, comprising: S1. Read the skirt number, dielectric section and electrical measurement position of the target UHV porcelain insulator. Divide the dielectric section into glaze slurry section, cleaning liquid section and pre-applied cleaning liquid section. Generate a leakage location map according to the pointing relationship from the glaze slurry section to the glaze surface return line position, the cleaning liquid section to the skirt root position, and the pre-applied cleaning liquid section to the appliqué edge position. This implementation method revolves around the leakage location map construction process. First, the structural cross-section of the target UHV porcelain insulator is converted into position information that can receive electrical measurements. Then, the contact trajectory of glaze slurry, cleaning fluid, and pre-applied cleaning fluid on the porcelain surface is converted into dielectric segments. Finally, a directional leakage location map is generated according to the arc length relationship between the dielectric segments and the electrical measurement positions, so that the subsequent trace detection results can fall on the glaze return line position, the umbrella skirt root position, or the adhesive edge position, rather than staying on the detection value of the dielectric itself. The implementation process includes the following steps: In S1-1, the axial profile data of the target UHV porcelain insulator is recorded using a two-dimensional point series under the same coordinate reference, with the coordinate unit of each point series being millimeters. The outer edge line of the umbrella skirt is formed by the outer periphery contour point series of the umbrella skirt. The upper surface line and lower surface line of the umbrella skirt are formed by the surface point series near the two sides of the axis under the same umbrella skirt number. The outer edge line of the hardware is formed by the contour point series of the hardware end. The glue filling edge line is formed by the contact boundary point series between the glue material and the porcelain end. The glaze layer coverage state is read along the sampling line of the umbrella skirt surface. The turning points from glazed to unglazed are connected in circumferential order to obtain the glaze flow termination line, and the glaze flow termination line is recorded as the glaze layer coverage state. The return flow line position; the broken line where the upper or lower surface line of the umbrella skirt intersects with the outer contour of the ceramic core is recorded as the umbrella skirt root position. When the contact trajectory is on the upper surface, the upper surface intersection broken line is taken; when the contact trajectory is on the lower surface, the lower surface intersection broken line is taken. The broken line where the glue filling edge line intersects with the outer contour of the ceramic end is recorded as the glue edge position. Each electrical test position is recorded with the product number, umbrella skirt number, electrical test position type, starting arc length, ending arc length, and sampling line number. The surface sampling line that is a single sampling position from the glaze return flow line position, umbrella skirt root position, or glue edge position and does not overlap with the arc length range of the leakage position is recorded as the adjacent surface position. The adjacent surface positions are only used for subsequent corroborative comparison, thereby generating the electrical test position table. In S1-2, the medium segment is generated from the medium contact record. The medium contact record includes at least the product number, medium type, flow skirt number, contact trajectory number, contact start arc length, contact end arc length, medium entry time, and medium exit time. When the same medium has a broken contact trajectory under the same skirt number, it is split into multiple medium segments according to the contact trajectory number. When the medium spans multiple skirt numbers, it is split into multiple medium segments according to the skirt number. The arc length is recorded with the skirt root position as zero point, along the upper surface line of the skirt towards the outer edge line of the skirt as the positive arc length, and along the lower surface line of the skirt towards the outer edge line of the skirt as the negative arc length. The contact start arc length is generated from the contact trajectory. The arc length of the curved surface from the starting point to the root of the umbrella skirt is obtained by combining the direction of the surface. The arc length of the contact endpoint is obtained by combining the arc length of the curved surface from the end of the contact trajectory to the root of the umbrella skirt with the direction of the surface. When the medium type is glaze slurry and the arc length of the contact endpoint falls between the starting arc length and the ending arc length of the glaze return line, the medium segment is recorded as the glaze slurry segment. When the medium type is cleaning fluid and the arc lengths of the contact starting point and the contact endpoint are located on both sides of the zero point of the umbrella skirt root, the medium segment is recorded as the cleaning fluid segment. When the medium type is pre-applied cleaning fluid and the arc length of the contact endpoint falls between the starting arc length and the ending arc length of the appliqué edge, the medium segment is recorded as the pre-applied cleaning fluid segment. This generates the medium segmentation table. In S1-3, first, read the glaze return line position, umbrella skirt root position, glued edge position, and adjacent surface position in the electrical measurement position table according to the product number and umbrella skirt number. Then, read the glaze slurry segment, cleaning fluid segment, and pre-glue cleaning fluid segment under the same product number and umbrella skirt number in the media segmentation table. For the glaze slurry segment, take the glaze slurry segment as the starting node, and take the glaze return line position under the same umbrella skirt number whose arc length range includes the contact endpoint arc length as the endpoint node, and write it to the pointing edge from the glaze slurry segment to the glaze return line position. For the cleaning fluid segment, take the cleaning fluid segment as the starting node, and take the glaze return line position located between the contact start arc length and the contact endpoint arc length under the same umbrella skirt number. The umbrella skirt root position is used as the endpoint node, and the pointing edge from the cleaning fluid segment to the umbrella skirt root position is written. For the pre-applied cleaning fluid segment, the pre-applied cleaning fluid segment is used as the starting node, and the applied edge position with an arc length range including the contact endpoint arc length under the same umbrella skirt number is used as the endpoint node, and the pointing edge from the pre-applied cleaning fluid segment to the applied edge position is written. The adjacent surface positions are connected to the adjacent glaze return line positions, umbrella skirt root positions, or applied edge positions under the same umbrella skirt number according to the adjacent sampling position relationship in the electrical measurement position table. The adjacent surface positions are not used as the endpoint nodes of the medium segment. The leakage position map is generated by the medium segment node, the leakage position node, the adjacent surface position node, and the corresponding pointing edge. Through the above processing, the structural position, medium contact trajectory, and subsequent electrical measurement position of the target UHV porcelain insulator are converted into the same leakage location map. The glaze slurry section, cleaning liquid section, and pre-applied cleaning liquid section can all be written into the specific leakage location according to the arc length range, while the adjacent surface position is retained as the subsequent counter-evidence location. This processing enables the subsequent residual point projection, retention point identification, and candidate path search to have fixed node and edge relationships, reducing the situation of mistakenly writing surface-passing medium residues into the leakage path. In practical applications: Taking an ultra-high voltage porcelain insulator with multiple skirts as an example, the production line first extracts axial profile data from the 3D scanning model and generates glaze return line positions, skirt root positions, adhesive edge positions, and adjacent surface positions according to the skirt numbers; when the glaze slurry flows to the vicinity of the outer edge of the sixth skirt, the arc length of the contact endpoint falls within the arc length range of the glaze return line position of the sixth skirt, and the system records this medium segment as the glaze slurry segment and connects it to the glaze return line position of the sixth skirt; when the cleaning fluid flows from the upper surface of the sixth skirt through the skirt root position and enters the lower surface, the arc length of the contact start point and the arc length of the contact end point are located on both sides of the zero point of the skirt root position, and the system records this medium segment as the cleaning fluid segment and connects it to the root position of the sixth skirt; when the cleaning fluid terminates at the end adhesive boundary before adhesive bonding, the system connects this medium segment to the adhesive edge position, finally forming a leakage bit map that can be used for subsequent trace detection and electrical parameter determination.

[0018] S2. Perform double-end trace detection on the medium segment, calculate the ion response increment and conductivity response increment of the same medium segment, record the medium segments with the same sign and non-zero as residual points, and write the residual points into the leakage position pointed to by the same medium segment through arc length projection calculation to generate residual projection record. This implementation focuses on the process of locating trace detection results of the dielectric segment. First, samples are taken at both ends of the dielectric entering and leaving the porcelain insulator surface. Then, a two-end differential algorithm is used to eliminate the difference in dielectric background. Subsequently, a same-sign rank selection algorithm is used to select the dominant ion response with the same direction of conductivity change from multiple ion channels. Finally, the residual points are written into the leakage bit map according to the dielectric type and arc length relationship, so that the residual points are not directly regarded as quality anomalies, but are first located at specific leakage locations. The target ion type code, electrode output unit, dielectric segment number, flow skirt number, and arc length field are all written with the record for subsequent stagnation point identification and candidate path search. This implementation process includes the following steps: In S2-1, for each dielectric segment, the system reads the dielectric entry time, dielectric exit time, dielectric type, and flow skirt number using the dielectric segment number as an index. An inlet sample is obtained before the dielectric enters the porcelain insulator surface, and an outlet sample is obtained after the dielectric leaves the porcelain insulator surface. Both inlet and outlet samples are recorded with the dielectric segment number, sampling time, sample type, dielectric type, and flow skirt number. The ion-selective electrode outputs detection values ​​according to the target ion species, with the target ion species code recorded as the ion channel number. For example, sodium ion channels, potassium ion channels, chloride ion channels, and sulfate ion channels use different codes. The millivolt value output by the ion-selective electrode is recorded as the ion electrode potential value, and the micro-Siemens per centimeter value output by the conductivity electrode is recorded as the conductivity value. Inlet and outlet samples are stored in pairs under the same dielectric segment number, and the inlet sample ion electrode potential value and the outlet sample ion electrode potential value are recorded separately under the same ion channel number, thus generating a segment sample record. In S2-2, the system uses a two-end differential algorithm to perform differential calculations based on segment sample records. First, it reads the ion electrode potential values ​​of the outlet sample and the inlet sample according to the medium segment number and ion channel number. Then, it subtracts the ion electrode potential value of the inlet sample from the outlet sample to obtain the ion channel increment. The conductivity response increment under the same medium segment is obtained by subtracting the conductivity value of the inlet sample from the conductivity value of the outlet sample. The ion channel increment retains the ion channel number, increment sign, and increment absolute value, while the conductivity response increment retains the increment sign and increment value. An increment greater than zero is written as positive, less than zero as negative, and equal to zero as zero. The segment differential record includes at least the medium segment number, ion channel number, ion channel increment, conductivity response increment, medium type, flow skirt number, contact start arc length, and contact end arc length, so that subsequent same-sign rank selection and arc length projection can continue to be performed along the same medium segment. In S2-3, the system reads the segment difference record using the same-sign rank selection algorithm. First, it deletes ion channel increments with the opposite sign to the conductivity response increment, and then deletes ion channel increments with the sign of zero. The retained ion channel increments are arranged in descending order of absolute value, with the first-ranked ion channel increment written as the ion response increment. If multiple retained ion channel increments have the same absolute value, the increment of the first ion channel in the ion channel table is taken as the ion response increment. If no retained ion channel increment exists, no residual point is generated in the dielectric segment. When both the ion response increment and the conductivity response increment are non-zero, the dielectric segment is recorded as a residual point. The residual point record includes the dielectric segment number, dielectric type, flow skirt number, contact start arc length, contact end arc length, ion channel number, ion response increment, and conductivity response increment. Otherwise, the dielectric segment only retains the segment difference record and does not enter the residual projection. In S2-4, the system uses an arc-length projection algorithm to write data based on the residual point and the leak location map. First, according to the skirt number flowing through the residual point, the system reads the glaze return line position, skirt root position, and adhesive edge position under the same skirt number from the leak location map. Then, it selects the writing rule according to the medium type in the residual point. When the medium type is glaze slurry, the system reads the starting and ending arc lengths of the glaze return line position. If the arc length of the contact endpoint is not less than the starting arc length and not greater than the ending arc length, the residual point is written to that glaze return line position. When the medium type is cleaning fluid, the system reads the starting and ending arc lengths of the skirt root position. The arc length range of the skirt root position... When the residual point is located within the closed interval formed by the arc length of the contact start point and the arc length of the contact end point, the residual point is written to the root position of the umbrella skirt. When the medium type is the pre-applied cleaning liquid section, the system reads the starting arc length and ending arc length of the appliqué edge. When the arc length of the contact end point falls within the range of the arc length, the residual point is written to the appliqué edge. The residual projection record is written with the product number, medium section number, residual point number, umbrella skirt number, leakage position type, leakage position arc length range, ion response increment, conductivity response increment, and projection writing rule number. Residual points that do not meet the corresponding writing rules are retained as unprojected residual records and are not included in the subsequent candidate path search. Through the above processing, the trace detection results of the medium segment are transformed from the original electrical measurement data of the inlet and outlet samples into residual projection records with the medium segment, ion channel, and leakage site location. The dual-end difference algorithm is used to eliminate the difference in the medium background, the same-sign rank selection algorithm is used to retain the ion response with the same direction as the change in conductivity, and the arc length projection algorithm is used to press the residual point into the glaze return line position, umbrella skirt root position, or glue edge position. This process allows the trace detection results to first form residual points, and then write them into the specific leakage position through the leakage position map, providing a positional basis for the identification of the retention point in the subsequent three-time electrical measurement values. In practical applications: When a certain glaze slurry flows through the sixth umbrella skirt, the sodium ion channel potential of the inlet sample is 20 millivolts, the sodium ion channel potential of the outlet sample is 35 millivolts, the conductivity of the inlet sample is 120 microsiemens per centimeter, and the conductivity of the outlet sample is 150 microsiemens per centimeter. In this case, both the sodium ion channel increment and the conductivity response increment are positive. If the sodium ion channel increment ranks first among ion channels of the same sign, the system records this glaze slurry segment as a residual point. Based on the relationship that the arc length of the contact endpoint of the sixth umbrella skirt falls within the arc length range of the glaze surface return line position, the residual point is written into the glaze surface return line position of the sixth umbrella skirt, forming a residual projection record.

[0019] S3. Read the three-time electrical measurement values ​​of the leak location and the adjacent surface location in the residual projection record, calculate the contact difference and recovery difference, and record the records with the same contact difference and recovery difference at the leak location and the records with no recovery difference at the adjacent surface location as retention points. This implementation method revolves around the process of generating retention points. It takes the residual points and leakage locations in the residual projection record, firstly obtains three-time electrical measurement values ​​at the leakage location and the adjacent surface location, then obtains the contact difference and recovery difference through the three-time difference algorithm, then uses the three-term symbol parallel algorithm to determine whether there are residual electrical measurement residues at the leakage location, and finally uses the adjacent position proof by contradiction algorithm to exclude surface residues in the same direction, so that the retention point only falls on the leakage location and not the adjacent surface location. The implementation process includes the following steps: In S3-1, the system uses the residual projection record as the entry point, reads the residual point number, product number, flow skirt number, leakage location type, and leakage location arc length range, and retrieves the adjacent surface location connected to the leakage location in the leakage location map; for pre-contact measurement, the electrical test record with the earliest time before the medium entry time is taken; for post-contact measurement, the electrical test record with the latest time after the medium exit time is taken; for recovery measurement, the electrical test record with the latest time after the drying recovery record ends is taken; for leakage location and adjacent surface location, the local capacitance picofarad value, low-frequency impedance amplitude ohm value, and insulation resistance megohm value are read respectively; the electrical test record is written with the product number, flow skirt number, electrical test location type, electrical test location arc length range, measurement time type, and three measured values; if there are multiple electrical test records of the same measurement time type, the electrical test record with the earliest interval value between the detection time and the contact time or recovery end time is taken to generate the location test record; In S3-2, the system uses a three-time difference algorithm based on the position measurement record to process the three-time electrical measurement values ​​under the same product number, the same umbrella skirt number, the same electrical measurement position type, and the same electrical measurement position arc length range. For the local capacitance picofarad value, the capacitance contact difference is obtained by subtracting the measurement value before contact from the value measured after contact, and the capacitance recovery difference is obtained by subtracting the measurement value before contact from the value measured after recovery. For the low-frequency impedance amplitude ohm value, the impedance contact difference is obtained by subtracting the measurement value before contact from the value measured after contact, and the impedance recovery difference is obtained by subtracting the measurement value before contact from the value measured after recovery. For the insulation resistance megohm value, the insulation resistance contact difference is obtained by subtracting the measurement value before contact from the value measured after contact, and the insulation resistance recovery difference is obtained by subtracting the measurement value before contact from the value measured after recovery. The position difference record includes the electrical measurement position type, electrical measurement position arc length range, measurement item name, contact difference, recovery difference, and difference sign. A difference greater than zero is written as a positive sign, less than zero as a negative sign, and equal to zero as a zero sign. In S3-3, the system reads only the position difference record of the leakage position using a three-term parallel algorithm, and judges the local capacitance picofarad value, low-frequency impedance amplitude ohm value, and insulation resistance megohm value as a measurement item in parallel. For any measurement item, if the contact difference is not zero, the recovery difference is not zero, and the signs of the contact difference and recovery difference are the same, then the measurement item is written as a leakage retention item. Positive contact difference and positive recovery difference of local capacitance picofarad value indicate that the capacitance increases after contact and the increase is retained after recovery. Negative contact difference and negative recovery difference of insulation resistance megohm value indicate that the insulation resistance decreases after contact and the decrease is retained after recovery. The low-frequency impedance amplitude ohm value is written according to the sign relationship of the measurement items with the same name. The leakage retention record is written with the leakage position number, measurement item name, contact difference sign, recovery difference sign, and number of leakage retention items. In S3-4, the system uses the adjacent position counter-proof algorithm to compare the leakage retention record and the position difference record of adjacent surface positions item by item. For each leakage retention item, the system first reads the measurement item name and recovery difference symbol of the leakage retention item, and then reads the recovery difference symbol of the measurement item with the same name in all adjacent surface positions connected to the same leakage position. If any adjacent surface position has a recovery difference with the same name and the same number, the leakage retention item is written into the adjacent position same number record. If no adjacent surface positions have a recovery difference with the same name and the same number, the leakage retention item is written into the adjacent position counter-proof pass record. When at least one leakage retention item in the leakage position forms an adjacent position counter-proof pass record, the system records the leakage position as a retention point and retains the original leakage position number, residual point number, measurement item name and recovery difference symbol in the retention point. Otherwise, the leakage position is not written as a retention point. Through the above processing, the leakage points in the residual projection record do not directly enter the subsequent candidate paths due to a single electrical measurement fluctuation. Instead, they first undergo four layers of processing: contact difference, recovery difference, leakage retention item, and adjacent position counter-evidence. The leakage point must undergo an electrical measurement change after contact and retain the same direction of change after recovery. At the same time, adjacent surface positions must not show the same name and the same sign of recovery difference before it is recorded as a retention point. This processing separates the surface diffusion residue from the residual retention within the leakage point, providing the end position for the subsequent non-return map search. In practical applications: A residual point of a certain glaze slurry segment has been written into the glaze return line position of the sixth umbrella skirt. The system reads the three-time electrical measurement values ​​of the glaze return line position and the adjacent surface position. If the local capacitance of the glaze return line position increases after contact relative to before contact, and still increases after recovery relative to before contact, while the recovery difference of the local capacitance of the adjacent surface position is zero or negative, then the local capacitance of the glaze return line position is written as a leakage retention item, and the glaze return line position is recorded as a retention point. If the adjacent surface position also has a recovery difference of the same sign of local capacitance, then this change is classified as surface same-direction residue, and the glaze return line position is not recorded as a retention point.

[0020] S4. Using the residual point as the starting point and the stagnation point as the ending point, expand the path along the pointing relationship of the leakage bit map according to the non-return graph search algorithm. Stop the path that enters the adjacent surface bit, returns to the leaked bit or has no subsequent leaked bit and does not reach the stagnation point, and generate candidate paths. This implementation method revolves around the candidate path generation process, taking into account residual projection records, leakage bitmaps, and stagnation points. The residual points are used as the search starting point, and the leakage bits written to the residual points are used as the first leakage bits. Then, the process is extended backward according to the pointing relationship in the leakage bitmap. The extension process only allows the entry of leakage bits that have not been written into the path string. Adjacent surface bits only participate in the proof of contradiction and do not participate in the path arrival. Leakage bits that have already been written into the path string will not enter the same path string again to avoid path loops causing false arrivals between residual points and stagnation points. The implementation process includes the following steps: In S4-1, the system reads the residual point number, product number, flow skirt number, medium segment number, leakage position type, and leakage position arc length range from the residual projection record, and generates a path start field using the residual point number; the leakage position written by the residual point retains the original leakage position number and is recorded as the first leakage position; the system generates an initial path string according to the writing order of "path start, first leakage position", and the initial path string must at least write the path string number, the path start residual point number, the first leakage position number, the current tail leakage position number, and the writing sequence number; among them, the current tail leakage position number is initially taken as the first leakage position number, and the writing sequence number increments from the path start; if multiple leakage positions are written to the same residual point, each leakage position generates an initial path string separately, and multiple leakage positions are not merged into the same path string; In S4-2, the system reads the outgoing edges of the current tail leak position in the leak bitmap based on the initial path string using a non-backward graph search algorithm, and records the node pointed to by the outgoing edge as the pointing node. The node type of the pointing node is read from the node table of the leak bitmap, and the node type includes leak position and adjacent surface position. If the pointing node is a leak position and the leak position number does not appear in the current path string, then the pointing node is written as a position to be traversed. If the pointing node is an adjacent surface position, then the branch is written as a stopped branch. If the pointing node is a leak position that has already been written into the current path string, then the branch is also written as a stopped branch. The stopped branch writes the branch starting leak position number, the branch ending node number and the reason for stopping. The position to be traversed writes the position to be traversed number, the source path string number and the source tail leak position number, which are used for the next round of path expansion. In S4-3, for a position to be traversed under the same current path string, the system copies the current path string one by one according to the writing order of the positions to be traversed in the leakage bit map side table, and appends each position to the end of the copied path string to form a new path string. The new path string synchronously updates the leakage bit number and writing order number at the end of the current path string. After appending, the system reads the leakage bit number at the end of the new path string and compares it with the leakage bit number carried by the stagnation point. If they are the same, the new path string is recorded as the arrival path string, and the subsequent expansion of the arrival path string is stopped. If they are different, the system continues to read the node pointed to by the leakage bit at the end of the new path string in the leakage bit map, and repeats the generation of positions to be traversed, the writing of the stopped branch, and the appending of the path string. The stagnation point retains the original leakage bit number, so the comparison object is the leakage bit number, and the measurement item name or difference symbol is not used as the basis for arrival. In S4-4, if the waiting position of a certain path is empty and the leakage position number at the end of the path is different from the leakage position number carried by the stagnation point, the system stops the expansion of the path and writes the path to the unreached record; if the path has been recorded as a reached path, the system reads the residual point number at the beginning of the path and the leakage position number at the end of the path. If the residual point number at the beginning of the path is the same as the residual point and the leakage position number at the end of the path is the same as the leakage position number carried by the stagnation point, the reached path is recorded as a candidate path; the candidate path is written with the candidate path number, the residual point number at the beginning of the path, the leakage position number sequence within the path, the stagnation point number at the end of the path, the number of stopped branches, and the candidate path writing order; the leakage position number sequence within the path is saved according to the writing order during the expansion process, providing the deletion object for subsequent point deletion, splitting and pruning; Through the above processing, the path between the residual point and the stagnation point is not generated according to the connectivity of the arbitrary graph, but is jointly limited by the pointing relationship of the leak bit graph, the adjacent surface bit prohibition rule, and the past leak bit prohibition return rule; the adjacent surface bit is reserved as a circumstantial position and does not participate in the path arrival; the leak bit that has been written into the path string no longer participates in the same path string expansion, so that the candidate path only expresses the unidirectional process of the residual point reaching the stagnation point along the direction of the leak bit; this process provides a recalculated path object for subsequent point deletion, pruning and splitting, reducing the situation where surface detour paths enter the placeholder judgment; In practical applications: When a residual point has been written into the sixth umbrella skirt glaze return line position, the system records the residual point as the beginning of the path, records the sixth umbrella skirt glaze return line position as the first leakage position, and generates an initial path string; if there is a pointing edge from the glaze return line position to the root position of the sixth umbrella skirt in the leakage position diagram, and the root position of the umbrella skirt has not been written into the path string, then the root position of the umbrella skirt is written as the position to be moved and appended to the end of the path string; if the next pointing node is an adjacent surface position, then the branch is written as the stopped branch; if the end of the path string eventually reaches the root position of the umbrella skirt that has been recorded as a stagnant point, then the path string is recorded as a candidate path for subsequent occupancy path determination.

[0021] S5. Recalculate the connection state of the candidate path by the deletion point splitting and pruning algorithm, perform residual substitution and lingering substitution proof on the splitting candidate path, record the splitting candidate path that was not substituted as a placeholder path, and output the effective leakage path location, re-inspection location and interception range. This implementation method revolves around the process of generating placeholder paths, taking into account candidate paths, residual points, leakage points, and stagnation points. First, the three-stage deletion and splitting algorithm is used to determine whether the key nodes in the candidate paths play a necessary role in the path from the residual point to the stagnation point. Then, the double substitution proof by contradiction algorithm is used to eliminate alternative interpretations formed by other residual points or other stagnation points. Finally, the lexicographical pruning and backwriting algorithm is used to select placeholder paths from the undetermined placeholder paths and transform the placeholder paths into effective leakage path locations, re-inspection locations, and interception ranges. This implementation process includes the following steps: In S5-1, the system reads the candidate road number, the number of the residual point at the beginning of the road, the sequence of leakage point numbers within the road, the number of the stagnant point at the end of the road, and the writing order of the candidate road based on the candidate road. It then writes the residual point at the beginning of the road, each leakage point within the road, and the stagnant point at the end of the road into the deletion object table. The stagnant point at the end of the road retains its original leakage point number and carries a stagnant mark, and is treated as a stagnant point deletion object, no longer being repeatedly written into the deletion object table as a regular leakage point. The three-stage deletion and splitting algorithm generates a copy of the leakage bitmap item by item according to the deletion object table: when deleting a residual point, if the beginning of the road does not exist, a disconnection position is directly written; when deleting a stagnant point, if the end of the road does not exist, a disconnection position is directly written; when deleting any leakage point, the leakage point and its incoming and outgoing edges are removed from the leakage bitmap copy, and then the non-return graph search is recalculated with the original residual point as the beginning of the road and the original stagnant point as the end of the road. If the search reaches the destination, a connectivity position is written; otherwise, a disconnection position is written. The system arranges the disconnection and connectivity positions according to the writing order of the deletion objects in the candidate roads, generating a deletion bit string. In S5-2, the system uses a major bit determination algorithm to read three types of deletion results based on the deletion point string: the first type is the disconnection bit after deleting the residual point, the second type is the disconnection bit after deleting the stagnant point, and the third type is the deletion status after deleting each leakage bit. If the deletion of the residual point results in a disconnection bit, the deletion of the stagnant point results in a disconnection bit, and at least one leakage bit in the candidate path results in a disconnection bit after deletion, then the system writes the candidate path into the split candidate path and writes the leakage bit that results in a disconnection bit as a split leakage bit. Otherwise, the candidate path is written into the non-occupied path. The non-occupied path records the candidate path number, the node number that does not form a split leakage bit, and the corresponding connected bit. It does not enter the effective leakage path position writing process and is only retained as the candidate path exclusion result. In S5-3, the system employs a double substitution proof-of-contrast algorithm for split candidate paths. First, it reads the current residual point, the current stagnation point, and the split leakage point. The current residual point is the starting residual point of the split candidate path, and the current stagnation point is the ending stagnation point. On the residual substitution side, all residual points within the same medium segment except the current residual point are used as residual substitution points. If a residual substitution point exists, the non-return graph search is recalculated using each residual substitution point as the starting point and the current stagnation point as the ending point. The search continues until a write connectivity point is reached; otherwise, a write disconnection point is reached. If no residual substitution point exists... If a replacement point is found, the residual replacement record is directly written to the disconnect position. For the lingering replacement side, the lingering points under the same cut leakage position, excluding this lingering point, are used as lingering replacement points. If a lingering replacement point exists, the non-return graph is recalculated with this residual point as the road start and each lingering replacement point as the road end. If the search reaches the connection position, it is written to the disconnect position; if it does not reach the disconnect position, it is written to the disconnect position. If no lingering replacement point exists, the lingering replacement record is directly written to the disconnect position. Both the residual replacement record and the lingering replacement record are written with the candidate road number, replacement point number, recalculated road start, recalculated road end, and connection status. In S5-4, the system reads the residual and stagnant substitution records of the candidate path to be severed. If any record is a connected bit, the candidate path to be severed is written into the substitution hit record, and the substitution hit record does not generate a valid leak path position; otherwise, the candidate path to be severed is written into the pending occupant path. The system uses a lexicographical pruning and write-back algorithm for the pending occupant path. First, it calculates the number of leaked bits, which is the number of leaked bits in the leaked bit number sequence within the path; then it calculates the number of disconnected bits, which is the number of deleted point objects with disconnected bits in the deleted point bit string; then, it sorts the pending occupant paths in ascending order of the number of leaked bits, descending order of the number of disconnected bits, and ascending order of the candidate path writing order. The system assigns the first position in the sequence as the placeholder path. The system writes the broken leakage points within the placeholder path as the effective leakage path location. The effective leakage path location is then written with the product number, skirt number, leakage point type, leakage point arc length range, placeholder path number, and residual point number. The skirt number, electrical testing point, and adjacent surface points to which the effective leakage path location belongs are written as the re-inspection location. The set of product numbers within the same testing batch that have the same medium segment or the same medium source identifier and flow through the same skirt number is written as the interception range. If the record contains only a single target UHV porcelain insulator, the interception range is written as the effective leakage path location of that target UHV porcelain insulator. Through the above processing, candidate paths must simultaneously satisfy node segmentation, no residual substitution, no lingering substitution, and lexicographical pruning retention before they can be written as placeholder paths. The three-stage segmentation algorithm is used to confirm the necessary role of the leak location in the path from the residual point to the lingering point. The double substitution proof algorithm is used to exclude alternative arrivals formed by other residual points or other lingering points under the same leak location within the same medium segment. The lexicographical pruning and write-back algorithm is used to select placeholder paths with fewer nodes, stronger disconnection effects, and earlier writing order from multiple undetermined placeholder paths. This processing ensures that the effective leak path location is not directly obtained from a single trace detection result or a single electrical measurement lingering result, but is jointly limited by the segmentation state of the candidate path and the substitution proof. In practical applications: a candidate route consists of the residual point of the glaze slurry section of the sixth umbrella skirt, the glaze return line position of the sixth umbrella skirt, the root position of the sixth umbrella skirt, and the stagnant point at the root position of the sixth umbrella skirt. After deleting the glaze return line position, the system recalculates the non-return map search. If the residual point cannot reach the stagnant point at the root position of the umbrella skirt, the glaze return line position is written as the cut leakage position. Then, other residual points in the same glaze slurry section and other stagnant points under the same umbrella skirt root position are searched. If both sides are recalculated as disconnected positions, the candidate route enters the pending occupancy route. After lexicographical pruning, the candidate route is ranked first. The system writes the glaze return line position of the sixth umbrella skirt as the effective leakage path position, and writes the glaze return line position of the sixth umbrella skirt and the adjacent surface positions into the re-inspection position. The set of product numbers covered by the same glaze slurry section is written into the interception range.

[0022] Working principle: This scheme first converts the sheds, glaze return lines, shed roots, and adhesive boundaries of the UHV porcelain insulator into a leakage location map. Then, trace detection is performed on the glaze slurry, cleaning fluid, and pre-adhesive cleaning fluid. The difference between the inlet and outlet samples is used to determine whether there are residual changes in the medium section, and the residual points are written into the specific leakage location according to the arc length relationship. Subsequently, the capacitance, low-frequency impedance, and insulation resistance of the leakage location and the adjacent surface locations before contact, after contact, and after recovery are read to determine whether there are electrical changes that are still retained after recovery at the leakage location, and whether the positions of adjacent surfaces do not change in the same direction. Finally, the path from the residual point to the retention point is searched along the leakage location map. Then, the path must pass through a certain leakage location by deleting points, recalculating, and substituting to confirm whether the path must pass through a certain leakage location, thereby determining whether the trace residue truly occupies the effective leakage path, and outputting the re-inspection location and interception range. In practical applications, for example, when a batch of UHV porcelain insulators passes through a glaze slurry section, the system first samples before and after the glaze slurry enters the porcelain surface. After completing trace detection, it is found that the ion response and conductivity response of the outlet sample change in the same direction, so the glaze slurry section is recorded as a residual point. If the arc length of the contact endpoint of the glaze slurry section falls on the glaze return line position of the sixth umbrella skirt, the system writes the residual point at that position. Subsequent electrical tests show that the capacitance of the glaze return line position of the sixth umbrella skirt increases after contact and there is still a residual change after recovery, while the adjacent surface position does not recover in the same direction. The system records this position as a stagnation point. After confirming through path search and point deletion recalculation that the residual point cannot reach the stagnation point after deleting the glaze return line position, and there are no other residual points or stagnation points that can replace the explanation, the glaze return line position of the sixth umbrella skirt is determined as the effective leakage path position, and based on this, re-inspection and interception of related products covered by the same glaze slurry section are arranged.

[0023] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A fully intelligent detection and quality control process for ultra-high voltage porcelain insulators, characterized in that, include: S1. Read the skirt number, dielectric section and electrical measurement position of the target UHV porcelain insulator. Divide the dielectric section into glaze slurry section, cleaning liquid section and pre-applied cleaning liquid section. Generate a leakage location map according to the pointing relationship from the glaze slurry section to the glaze surface return line position, the cleaning liquid section to the skirt root position, and the pre-applied cleaning liquid section to the appliqué edge position. S2. Perform double-end trace detection on the medium segment, calculate the ion response increment and conductivity response increment of the same medium segment, record the medium segments with the same sign and non-zero as residual points, and write the residual points into the leakage position pointed to by the same medium segment through arc length projection calculation to generate residual projection record. S3. Read the three-time electrical measurement values ​​of the leak location and the adjacent surface location in the residual projection record, calculate the contact difference and recovery difference, and record the records where the leak location has the same contact difference and the same recovery difference, and the adjacent surface location does not have the same recovery difference as the retention point. S4. Using the residual point as the starting point and the stagnation point as the ending point, expand the path along the pointing relationship of the leakage bit map according to the non-return graph search algorithm. Stop the path that enters the adjacent surface bit, returns to the leaked bit or has no subsequent leaked bit and does not reach the stagnation point, and generate candidate paths. S5. Recalculate the connection state of the candidate path by the deletion point splitting and pruning algorithm, perform residual substitution and lingering substitution proof on the splitting candidate path, record the splitting candidate path that was not substituted as a placeholder path, and output the effective leakage path location, re-inspection location and interception range.

2. The intelligent detection and quality control process for ultra-high voltage porcelain insulators according to claim 1, characterized in that: S1 includes: S1-1. Read the outer edge line of the umbrella skirt, the upper surface line of the umbrella skirt, the lower surface line of the umbrella skirt, the outer edge line of the fitting, and the glue filling edge line from the axial profile data of the target UHV porcelain insulator. Record the glaze flow termination line as the glaze return line position, the intersection line of the upper surface line of the umbrella skirt and the outer contour of the porcelain core as the umbrella skirt root position, the intersection line of the glue filling edge line and the outer contour of the porcelain end as the glue edge position, and record the surface sampling line of a single adjacent sampling position from the glaze return line position, the umbrella skirt root position, or the glue edge position as the adjacent surface position, and generate an electrical measurement position table. S1-2. For the media segment, read the media type, the umbrella skirt number, the arc length of the contact start point, and the arc length of the contact end point; record the media segment whose media type is glaze slurry and whose arc length of the contact end point falls within the arc length range of the glaze surface return line as the glaze slurry segment; record the media segment whose media type is cleaning liquid and whose arc lengths of the contact start point and the arc length of the contact end point are located on both sides of the arc length of the umbrella skirt root as the cleaning liquid segment; record the media segment whose media type is pre-applied cleaning liquid and whose arc length of the contact end point falls within the arc length range of the appliqué edge as the pre-applied cleaning liquid segment, and generate the media segment table; S1-3. Based on the electrical measurement position table and the medium segmentation table, connect the glaze slurry segment to the glaze return line position under the same umbrella skirt number, connect the cleaning liquid segment to the umbrella skirt root position under the same umbrella skirt number, connect the pre-applied cleaning liquid segment to the applicator edge position under the same umbrella skirt number, and connect each adjacent surface position to the adjacent glaze return line position, umbrella skirt root position, or applicator edge position under the same umbrella skirt number to generate a leak position map.

3. The intelligent detection and quality control process for ultra-high voltage porcelain insulators according to claim 2, characterized in that: S2 includes: S2-1. For each dielectric segment, take an inlet sample before the dielectric enters the surface of the porcelain insulator and an outlet sample after the dielectric leaves the surface of the porcelain insulator. Encode the target ion type of the ion-selective electrode as the ion channel number, record the millivolt value output by the ion-selective electrode as the ion electrode potential value, and record the micro-Siemens per centimeter value output by the conductivity electrode as the conductivity value, and generate a segment sample record. S2-2. Based on segment sample records, a two-end difference algorithm is used to subtract the ion electrode potential value of the inlet sample from the ion electrode potential value of the outlet sample under the same ion channel number to obtain the ion channel increment, and subtract the conductance value of the inlet sample from the conductance value of the outlet sample to obtain the conductance response increment, thus generating segment difference records.

4. The intelligent detection and quality control process for ultra-high voltage porcelain insulators according to claim 3, characterized in that: S2 further includes: S2-3. Using the same-sign rank selection algorithm, retain the ion channel increments with the same sign as the conductivity response increment, arrange them in descending order of the absolute value of the retained ion channel increments, write the ion channel increment at the top of the sort as the ion response increment, and record the medium segment where both the ion response increment and the conductivity response increment are not zero as the residual point. S2-4. Based on the residual point, medium type, flow through the umbrella skirt number, contact start arc length, contact end arc length, and leakage location map, the arc length projection algorithm is used. First, the leakage location under the same umbrella skirt number is selected according to the flow through the umbrella skirt number. Then, the residual point is written according to the medium type: the residual point of the glaze slurry section is written as the glaze return line position with an arc length range including the contact end arc length; the residual point of the cleaning liquid section is written as the umbrella skirt root position with an arc length range between the contact start arc length and the contact end arc length; and the residual point of the pre-gluing cleaning liquid section is written as the gluing edge position with an arc length range including the contact end arc length. This generates a residual projection record.

5. The intelligent detection and quality control process for ultra-high voltage porcelain insulators according to claim 4, characterized in that: S3 includes: S3-1. Read the leakage position written by the residual point according to the residual projection record, and read the adjacent surface position connected to the leakage position from the leakage position map. Read the local capacitance picofarad value, low frequency impedance amplitude ohm value and insulation resistance megohm value of the leakage position and adjacent surface position respectively during the pre-contact test, post-contact test and recovery test, and generate the position test record. S3-2. Based on the position measurement record, a three-time difference algorithm is used to subtract the corresponding measured values ​​from the pre-contact measurement values ​​from the local capacitance picofarad value, low-frequency impedance amplitude ohm value, and insulation resistance megohm value measured after contact to obtain the contact difference value. The corresponding measured values ​​from the pre-contact measurement value are then subtracted from the corresponding measured values ​​during the recovery measurement to obtain the recovery difference value, thus generating the position difference record.

6. The intelligent detection and quality control process for ultra-high voltage porcelain insulators according to claim 5, characterized in that: S3 further includes: S3-3. Using the three-symbol parallel algorithm, read the position difference record of the leakage position, and write the measurement items in the local capacitance picofarad value, low frequency impedance amplitude ohm value and insulation resistance megohm value, where the contact difference and recovery difference are all non-zero and have the same sign, as leakage retention items, and bind the leakage retention items to the leakage position to generate leakage retention records. S3-4. Using the neighboring position proof algorithm, the recovery difference sign of the leakage retention item is compared with the recovery difference sign of the corresponding measurement item of the adjacent surface position. The leakage position of the adjacent surface position that does not have the same name and the same sign recovery difference is recorded as the retention point.

7. The intelligent detection and quality control process for ultra-high voltage porcelain insulators according to claim 6, characterized in that: S4 includes: S4-1. Read the leak bits written by the residual points according to the residual projection record, record the residual points as the road start, record the leak bits as the first leak bit, and generate the initial road string by the road start and the first leak bit in the writing order. S4-2. Based on the initial path string, read the pointing nodes of the leak bits at the end of the path string in the leak bit graph along the pointing relationship according to the non-backward graph search algorithm, write the leak bits that have not been written into the path string as the bits to be traversed, and write the adjacent surface bits and the branches that have been written into the path string as the branches to be stopped.

8. The intelligent detection and quality control process for ultra-high voltage porcelain insulators according to claim 7, characterized in that: S4 further includes: S4-3. For each position to be moved, add each position to the end of the path string and repeat the node pointed to by the leaked position at the end of the path string until the leaked position at the end of the path string is the same as the stagnation point, and generate the arrival path string. S4-4. Stop extending the path when the position to be moved is empty and the leakage point at the end of the path is different from the stagnation point. Record the path that arrives with the stagnation point at the beginning of the path and the stagnation point at the end of the path as a candidate path.

9. The intelligent detection and quality control process for ultra-high voltage porcelain insulators according to claim 8, characterized in that: S5 includes: S5-1. Based on the candidate path, a three-deletion splitting algorithm is adopted. The residual points, each leakage point and the stagnation point in the candidate path are respectively regarded as deletion objects. After removing one deletion object at a time, the non-return graph search is recalculated. If the residual point is disconnected from the stagnation point, the disconnection position is written. If the residual point reaches the stagnation point, the connection position is written, and the deletion point bit string is generated. S5-2. Based on the deletion point bit string, use the major bit determination algorithm. If the deletion of the leaked bit results in a disconnected bit, the deletion of the residual point results in a disconnected bit, and the deletion of the stagnant point results in a disconnected bit, then write the candidate path into the split candidate path; otherwise, write the candidate path into the non-occupied path.

10. The intelligent detection and quality control process for ultra-high voltage porcelain insulators according to claim 9, characterized in that, include: The S5 also includes: S5-3. For the candidate path of the split, the double substitution proof algorithm is adopted. The residual points other than the current residual point in the same medium segment are used as residual substitution points. The non-return graph search from the residual substitution point to the current stagnation point is recalculated and residual substitution records are generated. The stagnation points other than the current stagnation point under the same leakage position are used as stagnation substitution points. The non-return graph search from the current residual point to the stagnation substitution point is recalculated and stagnation substitution records are generated. S5-4. If the residual or delayed replacement record is a connected bit, the candidate path for severance is written into the replacement hit record. Otherwise, the candidate path for severance is written into the pending occupancy path. The lexicographical pruning and write-back algorithm is used to select the occupancy path in ascending order of the number of leakage bits, descending order of the number of disconnection bits, and ascending order of candidate path writing. The leakage bits in the occupancy path are written as the effective leakage path location. The umbrella skirt number and electrical measurement bit to which the effective leakage path location belongs are written as the re-inspection location. The target UHV porcelain insulator with the same medium segment as the residual point in the occupancy path is written as the interception range.

Images