A robot knowledge expression method and system based on structure recognition

Abstract

The application is suitable for the technical field of intelligent maintenance and robot control, and provides a robot knowledge expression method and system based on structure identification, which comprises the following steps: acquiring the local oil flow disturbance intensity of a current area to be cleaned, and analyzing whether the local oil flow disturbance intensity exceeds the local oil flow disturbance intensity corresponding to a node; if yes, setting the cleaning comprehensive index corresponding to the node as the reference cleaning comprehensive index used in this cleaning; generating a correction factor according to the difference between the current local oil flow disturbance intensity and the local oil flow disturbance intensity corresponding to the node, and correcting the preset complementary factor parameter according to the correction factor. The mechanism of the application balances the cleaning effect and the risk of recontamination without excessively increasing the energy consumption and mechanical load, significantly improves the intelligent decision level and environmental adaptability of the cleaning robot, and is especially suitable for high-precision maintenance operation in a complex oil-cooled transformer.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of intelligent maintenance and robot control, and particularly relates to a robot knowledge expression method and system based on structure identification. BACKGROUND

[0002] With the continuous development of intelligent maintenance and unmanned operation technology, cleaning robots are increasingly used in complex and closed environments such as power equipment, industrial pipelines and heat exchange systems. In particular, in oil-immersed power transformers, the internal structure includes a large number of flow channels, insulation support members and heat dissipation oil circuits, and the overall geometric shape is complex and the local oil flow state varies significantly. During long-term operation, factors such as oil aging, impurity deposition and local flow disorder can cause fouling and accumulation of contaminants on the surface of the insulation structure, thereby affecting the heat dissipation efficiency and electrical insulation performance. In order to ensure the safe and stable operation of the equipment, it is necessary to use a cleaning robot to carry out high-precision and low-risk internal cleaning operations.

[0003] The cleaning robot in the prior art usually relies on fixed parameter setting or experience model to perform cleaning tasks, and enhances the cleaning intensity by increasing the jet flow rate, pressure pulse frequency, etc. However, this method mainly controls based on static structure identification results, and does not fully consider the dynamic coupling relationship between the "structure feature-oil flow disturbance-pollution redistribution" in the power transformer. The knowledge expression layer lacks semantic understanding and reasoning ability for complex physical environments, and cannot adjust the decision according to the structure identification results and the real-time environmental state, resulting in that in the high oil flow disturbance area, the cleaning comprehensive index is too high to easily cause the migration and reattachment of contaminants in the non-target area, forming secondary pollution; and when the cleaning intensity is reduced to avoid re-pollution, there is no effective compensation mechanism, and the cleaning efficiency is significantly reduced, and the cleaning result is unstable.

[0004] Therefore, the existing cleaning robot still has significant limitations in the knowledge expression layer, and cannot realize the environment adaptive cleaning strategy based on structure identification. How to integrate the structure identification information and the local oil flow disturbance feature, and realize the dynamic optimization and regulation of the cleaning comprehensive index and the auxiliary cleaning parameters through knowledge expression, has become a key technical problem to improve the intelligent decision-making ability and environmental adaptability of the cleaning robot. SUMMARY

[0005] The present application relates to the technical field of intelligent maintenance and robot control, and particularly relates to a robot knowledge expression method and system based on structure identification.

[0006] The present application is implemented in the following manner. In a first aspect, the present application provides a robot knowledge expression method based on structure identification, which comprises:

[0007] Determine the structural features of the internal area to be cleaned of the power transformer that the cleaning robot should deal with, and obtain historical cleaning task data of the cleaning robot;

[0008] Analyze the historical cleaning task data to obtain a plurality of sample data that the cleaning robot faces when the structural features are consistent with the current area to be cleaned, but the cleaning comprehensive index is different;

[0009] Obtain the re-pollution index corresponding to each sample data after cleaning, and analyze the node at which the re-pollution index exceeds the preset threshold for the first time when the cleaning comprehensive index gradually increases, and determine the local oil flow disturbance intensity corresponding to the node;

[0010] Obtain the local oil flow disturbance intensity of the current area to be cleaned, and analyze whether it exceeds the local oil flow disturbance intensity corresponding to the node; if so, set the cleaning comprehensive index corresponding to the node as the reference cleaning comprehensive index adopted during the cleaning;

[0011] Generate a correction factor according to the difference between the current local oil flow disturbance intensity and the local oil flow disturbance intensity corresponding to the node, and correct the preset complementary factor parameter according to the correction factor.

[0012] As a further limitation of the technical scheme of the embodiment of the application, the structural features of the area to be cleaned include the geometric shape parameters, channel space distribution characteristics, surface roughness parameters and scaling conditions of the area to be cleaned.

[0013] As a further limitation of the technical scheme of the embodiment of the application, the specific cleaning comprehensive index refers to a composite parameter representing the cleaning effect intensity of the cleaning robot during cleaning, including at least one of the jet flow rate or the jet pressure pulse frequency or a combination of the two parameters.

[0014] As a further limitation of the technical scheme of the embodiment of the application, the specific cleaning comprehensive index adopted by the cleaning robot when facing different sample data consistent with the current structural features of the area to be cleaned is different, and the reason for the difference is that the local oil flow disturbance intensity of the cleaned area corresponding to each sample data is different.

[0015] As a further limitation of the technical scheme of the embodiment of the application, the step of obtaining the re-pollution index corresponding to each sample data after cleaning, and analyzing the node at which the re-pollution index exceeds the preset threshold for the first time when the cleaning comprehensive index gradually increases, and determining the local oil flow disturbance intensity corresponding to the node includes:

[0016] In turn, analyze each sample data to obtain the re-pollution index of the corresponding cleaning robot after completing the cleaning of the internal area of the power transformer,

[0017] The recontamination index is used to represent the degree of migration, adhesion or deposition of contaminants in non-target areas due to the increase of jet flow rate or jet pressure pulse frequency under the action of a specific cleaning comprehensive index;

[0018] Different sample data are sorted according to the cleaning comprehensive index from small to large, and the change trend of the recontamination index in the sequence is analyzed;

[0019] If there is a node that first exceeds the preset threshold in the change trend of the recontamination index, and the recontamination index after the node exceeds the preset threshold, the local oil flow disturbance intensity corresponding to the node is determined as the recontamination intensity threshold.

[0020] As a further limitation of the technical scheme of the embodiment of the application, the step of generating a correction factor according to the difference between the current local oil flow disturbance intensity and the local oil flow disturbance intensity corresponding to the node, and correcting the preset complementary factor parameter according to the correction factor comprises:

[0021] The improvement range of the current local oil flow disturbance intensity compared to the recontamination intensity threshold is calculated, and the improvement range is used as the correction factor;

[0022] The correction factor and the preset correction amplitude coefficient are multiplied and jointly act on the preset complementary factor parameter to improve the complementary factor parameter, so as to compensate for the decrease in cleaning efficiency caused by the decrease in the cleaning comprehensive index, and realize the dynamic balance of the knowledge expression layer to the cleaning efficiency and the recontamination risk.

[0023] As a further limitation of the technical scheme of the embodiment of the application, the preset complementary factor parameter is the cleaning oil temperature or the jet medium temperature, and the cleaning oil temperature or the jet medium temperature is increased to reduce the viscosity of the oil and enhance the sediment stripping capacity, so that the cleaning effect is stable under the condition of reducing the cleaning comprehensive index.

[0024] In a second aspect, based on the same inventive concept, the application also provides a robot knowledge expression system based on structure recognition, which comprises:

[0025] A structure recognition module is configured to determine the structural characteristics of the internal area to be cleaned of the power transformer that the cleaning robot should deal with, and obtain historical cleaning task data of the cleaning robot;

[0026] A sample analysis module is configured to analyze the historical cleaning task data, and obtain a plurality of sample data of the cleaning robot in the face of the same structural characteristics as the current area to be cleaned but different specific cleaning comprehensive indexes;

[0027] A recontamination identification module is configured to obtain a recontamination index corresponding to each sample data after cleaning, and analyze a node at which the recontamination index exceeds a preset threshold for the first time when the cleaning comprehensive index gradually increases, and determine a local oil flow disturbance intensity corresponding to the node;

[0028] A disturbance analysis module is configured to obtain the local oil flow disturbance intensity of the current cleaning area, and analyze whether the local oil flow disturbance intensity exceeds the local oil flow disturbance intensity corresponding to the node; if yes, the cleaning comprehensive index corresponding to the node is set as a reference cleaning comprehensive index adopted in the current cleaning;

[0029] A knowledge correction module is configured to generate a correction factor according to the difference between the current local oil flow disturbance intensity and the local oil flow disturbance intensity corresponding to the node, and correct the preset complementary factor parameter according to the correction factor.

[0030] As a further limitation of the technical scheme of the embodiment of the present application, the structural characteristics of the cleaning area include geometric parameters, channel space distribution characteristics, surface roughness parameters and fouling conditions of the cleaning area.

[0031] As a further limitation of the technical scheme of the embodiment of the present application, the specific cleaning comprehensive index refers to a composite parameter representing the cleaning intensity applied by the cleaning robot during the cleaning process, including at least one of the jet flow rate or the jet pressure pulse frequency or a combination of the two parameters.

[0032] Compared with the prior art, the present application has the following beneficial effects:

[0033] The present application introduces a dynamic adjustment mechanism based on the local oil flow disturbance intensity in the knowledge expression layer, which realizes the collaborative optimization of the cleaning comprehensive index and the preset complementary factor parameter. Unlike the prior art which only relies on increasing the jet flow rate or the pressure pulse frequency to enhance the cleaning intensity, the present application first determines the recontamination intensity threshold through structure recognition and historical data analysis, and limits the upper limit of the cleaning comprehensive index on this basis, thereby inhibiting the risk of uncontrollable recontamination from the source; then, a correction factor is generated according to the excess amplitude of the current disturbance intensity, and the cleaning oil temperature or the jet medium temperature is adaptively increased to realize compensatory enhancement of the cleaning efficiency. This mechanism balances the cleaning effect and the recontamination risk without excessively increasing energy consumption and mechanical load, significantly improves the intelligent decision-making level and environmental adaptability of the cleaning robot, and is especially suitable for high-precision maintenance operations in complex oil-cooled transformers. BRIEF DESCRIPTION OF DRAWINGS

[0034] Figure 1 A flowchart of the method provided by the embodiment of the present application is shown in the figure;

[0035] Figure 2A flow chart of a recontamination intensity threshold determination process in the method provided by the embodiment of the present application is shown.

[0036] Figure 3 A flow chart of a complementary factor parameter correction process in the method provided by the embodiment of the present application is shown.

[0037] Figure 4 An application architecture diagram of the system provided by the embodiment of the present application is shown. DETAILED DESCRIPTION

[0038] In order to make the objectives, technical solutions and advantages of the present application clearer, the present application is further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and should not be used to limit the present application.

[0039] Embodiment one:

[0040] Figure 1 A flow chart of the method provided by the embodiment of the present application is shown.

[0041] Specifically, a robot knowledge expression method based on structure recognition, which specifically comprises the following steps:

[0042] In step S100, the structural features of the internal area to be cleaned of the power transformer that the cleaning robot should deal with are determined, and the historical cleaning task data of the cleaning robot is obtained. The structural features of the internal area to be cleaned include the geometric shape parameters, channel space distribution features, surface roughness parameters and fouling conditions of the internal area to be cleaned.

[0043] In the embodiment of the present application, the cleaning robot is used to clean the internal area of the power transformer. The power transformer is a device that realizes power transmission and voltage conversion through electromagnetic induction principle, which is usually composed of a core, a winding, an oil tank, a cooling device and an insulation system, etc., wherein the oil tank is filled with transformer oil for cooling and insulation. Due to the phenomena of oil flow circulation, heat conduction and impurity deposition in the internal area of the transformer during long-term operation, the accumulation of pollutants such as oil sludge, metal particles, carbonized deposits or aging residues of insulating materials in some internal areas is easy to occur, which affects the heat dissipation efficiency and insulation performance, and thus needs to be cleaned regularly.

[0044] The internal area to be cleaned of the power transformer can include but is not limited to the deposition area at the bottom of the oil tank, the internal area of the cooling oil channel, the winding gap channel, the surface of the core clamping structure, the surface of the insulation support, the radiator channel, the inlet and outlet channel of the oil pump, the inner wall of the communication pipeline, etc. These areas have complex spatial distribution features, different geometric dimensions and non-uniform fluid disturbance conditions in structure, so that the attachment form, deposition thickness and recontamination trend of the pollutants have strong structure dependence.

[0045] The structural features of the region to be cleaned are used to characterize the identification results of the robot on the target region before performing the cleaning operation, and specifically include geometric parameters (such as surface curvature, spatial angle, channel width, depth, and opening size, etc.), channel space distribution features (including main flow channel direction, branch channel connection relationship, oil flow path distribution density, etc.), surface roughness parameters (such as surface micro-protrusion height, waviness, local unevenness, etc.), and fouling conditions (including pollutant types, thickness distribution, adhesion density, and local deposition morphology). In addition, thermal flow density parameters, local oil flow disturbance intensity distribution, temperature gradient distribution, insulation material type, material surface energy parameters, etc. related to the cleaning environment can also be further included, and these structural and physical features jointly affect the accessibility of the cleaning behavior, the difficulty of pollutant peeling, and the probability of recontamination.

[0046] The above identification of the structural features of the region to be cleaned belongs to the structural identification technology that can be directly adopted by the existing cleaning robot. Through laser ranging, three-dimensional vision, ultrasonic detection, or structured light scanning, etc. sensing means, combined with point cloud reconstruction or feature segmentation algorithm, the robot can realize the spatial modeling and parameter extraction of the internal structure of the power transformer, so as to obtain the structural feature data for subsequent knowledge expression processing.

[0047] The historical cleaning task data is used to provide learning samples for the cleaning robot based on past work experience, which is derived from the cleaning task records of the robot in multiple different transformers or different structural feature regions. This data can be automatically collected by the robot's own running record module, or can be uniformly stored by the maintenance management system. The historical cleaning task data specifically includes structural feature data of the cleaning task (including the shape, size, and material information of the target region), cleaning execution parameters (such as jet flow rate, jet pressure, pulse frequency, jet medium temperature, cleaning duration, etc.), cleaning effect evaluation data (such as the proportion of residual pollution on the surface after cleaning, local recontamination rate, oil turbidity change, etc.), and task execution environment data (including local oil temperature, flow velocity, vibration intensity, operating load, etc.). Through analysis of the historical cleaning task data, data support can be provided for the decision-making of the subsequent knowledge expression layer, so that the robot has self-adaptive decision-making ability when facing different structural features.

[0048] Further, the robot knowledge expression method based on structural identification further includes the following steps:

[0049] In step S200, the historical cleaning task data is analyzed to obtain several sample data of the cleaning robot facing the same structural features as the current region to be cleaned, but different specific cleaning comprehensive indicators. When the cleaning robot faces different sample data with the same structural features as the current region to be cleaned, the specific cleaning comprehensive indicators used are different, and the reason for the difference is that the local oil flow disturbance intensity of the cleaned region corresponding to each sample data is different.

[0050] The specific cleaning comprehensive index refers to a composite parameter representing the cleaning effect intensity of the cleaning robot in the cleaning process, including at least one of the jet flow rate or the jet pressure pulse frequency or a combination of both parameters.

[0051] In the embodiment of the present application, the local oil flow disturbance intensity is used to represent the flow stability and local shear effect intensity of the oil flow in the transformer in a specific area, which is a dynamic characteristic parameter determined by the oil flow channel structure, heat dissipation flow direction and obstacle distribution during the operation of the transformer. The parameter reflects the redistribution and adhesion trend of the pollutants in the local space, which is a physical characteristic specific to the internal environment of the power transformer and an important environmental factor that the cleaning robot must directly face and respond to during the execution of the task.

[0052] The local oil flow disturbance intensity can be obtained by a micro flow rate sensor, a pressure fluctuation sensor or an oil flow turbulence monitoring module installed on the robot body or a detection device. The system acquires the local oil flow velocity variation, pressure fluctuation frequency and vortex distribution data in real time, combines with the internal structure recognition model for numerical fitting or fluid simulation reconstruction, so as to obtain a quantitative index for representing the disturbance degree.

[0053] The jet flow rate and the jet pressure pulse frequency are the main control parameters of the cleaning robot during the cleaning process, which together determine the impact energy, shear strength and fluid stripping capacity of the jet medium on the target surface, and can directly reflect the comprehensive intensity of the cleaning effect. There is a significant corresponding relationship between the two and the local oil flow disturbance intensity: when the local oil flow disturbance intensity is high, the pollutants are easy to migrate after cleaning and reattach in the adjacent area. At this time, if a higher jet flow rate or pressure pulse frequency is still used, the excessive input of cleaning energy will intensify the oil flow disturbance and further induce the recontamination phenomenon.

[0054] The core research point of the present application is that the cleaning robot comprehensively considers the structure recognition result and the local oil flow disturbance intensity at the knowledge expression level, and determines the cleaning strategy parameters through the collaborative analysis of the two. Among them, the specific cleaning comprehensive index (composed of the jet flow rate and the jet pressure pulse frequency) is used to reflect the cleaning effect intensity, and the cleaning oil temperature or the jet medium temperature is used as the main auxiliary cleaning factor, which can reduce the oil viscosity, enhance the fluid shear force and the stripping capacity of the pollutants, so as to realize the effective removal of the deposited pollutants without relying on the high cleaning comprehensive index. In other words, the cleaning comprehensive index and the oil temperature parameter together constitute two types of decision variables of the knowledge expression layer, the former is used to directly control the cleaning effect intensity, and the latter realizes the indirect improvement of the cleaning energy efficiency through the thermophysical regulation, and the two work together to realize the comprehensive balance of the cleaning efficiency and the recontamination risk.

[0055] In the prior art, the cleaning robot is usually controlled mainly by a specific cleaning comprehensive index, i.e., knowledge understanding and parameter setting are mainly performed from the aspect of cleaning intensity. The advantage of this strategy is that when the cleaning comprehensive index is high, the cleaning efficiency is significantly improved, and the deposits can be quickly removed. However, high-intensity spraying will also bring high mechanical impact, and in areas where the local oil flow disturbance is large, it is easy to cause secondary suspension and migration of pollutants, thereby causing the problem of recontamination. At the same time, although the adjustment of oil temperature or medium temperature can improve the cleaning effect, it requires additional high power input, which is not economical for wireless operation or energy-limited robots (and the required temperature raising time is long).

[0056] The present inventor has found through in-depth research that when the local oil flow disturbance intensity is large, maintaining a high cleaning comprehensive index will result in a significant risk of recontamination. Therefore, at the knowledge expression understanding level, by identifying the disturbance intensity and dynamically adjusting the cleaning comprehensive index and complementary factors (such as cleaning oil temperature or spraying medium temperature), the balance control of cleaning efficiency and recontamination risk can be achieved under different disturbance conditions. In other words, the present application proposes a self-adaptive strategy of dynamically reducing the cleaning comprehensive index and reasonably increasing the oil temperature type complementary parameter at the knowledge expression level for areas with the same structural feature but different local oil flow disturbance intensities, thereby achieving effective inhibition of recontamination and continuous optimization of cleaning effect.

[0057] Further, the robot knowledge expression method based on structure recognition further comprises the following steps:

[0058] Step S300, obtaining the recontamination index after cleaning corresponding to each sample data, and analyzing the node at which the recontamination index first exceeds the preset threshold when the cleaning comprehensive index gradually increases, and determining the local oil flow disturbance intensity corresponding to the node.

[0059] Specifically, Figure 2 A flow chart of the recontamination intensity threshold determination process is shown.

[0060] The step of obtaining the recontamination index after cleaning corresponding to each sample data, and analyzing the node at which the recontamination index first exceeds the preset threshold when the cleaning comprehensive index gradually increases, and determining the local oil flow disturbance intensity corresponding to the node specifically comprises the following steps:

[0061] Step S301, sequentially analyzing each sample data to obtain the recontamination index of the corresponding cleaning robot after completing the internal cleaning of the power transformer, the recontamination index being used to represent the degree of contamination migration, adhesion or deposition of non-target areas caused by the increase of jet flow rate or jet pressure pulse frequency under the action of a specific cleaning comprehensive index;

[0062] Step S302, sort the different sample data according to the cleaning comprehensive index from small to large, and analyze the change trend of the recontamination index under the sequence;

[0063] Step S303, if there is a node that first exceeds the preset threshold in the change trend of the recontamination index, and the recontamination index after the node all exceeds the preset threshold, it is determined that the local oil flow disturbance intensity corresponding to the node is the recontamination intensity threshold.

[0064] In the embodiment of the application, the main purpose of step S300 is to determine the critical relationship between the cleaning comprehensive index and the recontamination risk. It can be understood that the recontamination phenomenon does not appear significantly at any cleaning intensity, but there is an acceptable tolerance interval, that is, when the cleaning comprehensive index is within a certain range, the recontamination degree after cleaning is in a controllable state, and only after exceeding a certain characteristic critical point, the recontamination index will be significantly increased. Therefore, by analyzing the corresponding relationship between the cleaning comprehensive index and the recontamination index in the sample data, the critical node can be determined, and the local oil flow disturbance intensity corresponding to the node is taken as the recontamination intensity threshold, thereby providing a scientific basis for the subsequent dynamic adjustment of the cleaning parameters.

[0065] In step S301, the recontamination index is obtained based on the historical cleaning data of the cleaning robot in the past tasks. Specifically, it can be collected by multiple types of sensors installed on the robot body and its operating environment, for example:

[0066] The image sensor or microscopic camera is used to identify the distribution of pollutants remaining on the target surface after cleaning, and the image difference algorithm or the surface residue detection model based on deep learning is used to quantify the pollution area ratio of the area before and after cleaning;

[0067] The particle concentration sensor or oil quality monitoring module is used to detect the suspended particle concentration and deposition rate in the oil after cleaning, so as to reflect the re-suspension and migration degree of pollutants;

[0068] The conductivity or infrared scattering sensor is used to detect the change of trace pollutants or impurity content in the oil, so as to reflect the recontamination trend.

[0069] The above detection results are normalized and feature extracted by the data processing unit in the cleaning robot control system, forming a comprehensive numerical index for representing the recontamination degree. Since these sensors and computing means are the conventional capabilities possessed by the prior art, the process of obtaining the recontamination index can be realized in the existing robot perception system. All corresponding data are recorded in the historical cleaning task database for subsequent extraction and model training of sample data.

[0070] In step S303, by sorting and analyzing the change trend of the recontamination index of different sample data with the cleaning comprehensive index, the system can identify the node at which the recontamination begins to increase significantly. When it is found that the recontamination index first exceeds the preset threshold and remains higher than the threshold thereafter, it indicates that the cleaning intensity of this node has exceeded the tolerance limit of the system environment. The local oil flow disturbance intensity corresponding to this node is determined as the recontamination intensity threshold, which means that the oil flow environment under this disturbance intensity will cause the risk of secondary migration and attachment of pollutants to rise significantly. This setting provides a key basis for judging whether the current cleaning area exceeds the controllable range in the subsequent steps and adjusting the cleaning comprehensive index and complementary factors accordingly, so that the knowledge expression layer realizes quantifiable balance control between cleaning intensity and recontamination risk.

[0071] Further, the robot knowledge expression method based on structure recognition further comprises the following steps:

[0072] Step S400, the local oil flow disturbance intensity of the current cleaning area is obtained, and whether it exceeds the local oil flow disturbance intensity corresponding to the node is analyzed; if so, the cleaning comprehensive index corresponding to the node is set as the reference cleaning comprehensive index adopted this time.

[0073] Step S500, a correction factor is generated according to the difference between the current local oil flow disturbance intensity and the local oil flow disturbance intensity corresponding to the node, and the preset complementary factor parameter is corrected according to the correction factor.

[0074] Specifically, Figure 3 A flow chart of the complementary factor parameter correction process is shown.

[0075] Among them, the correction factor is generated according to the difference between the current local oil flow disturbance intensity and the local oil flow disturbance intensity corresponding to the node, and the preset complementary factor parameter is corrected according to the correction factor, which specifically includes the following steps:

[0076] Step S501, the improvement amplitude of the current local oil flow disturbance intensity compared with the recontamination intensity threshold is calculated, and the improvement amplitude is taken as the correction factor;

[0077] Step S502, the correction factor and the preset correction amplitude coefficient are multiplied to jointly act on the preset complementary factor parameter to improve the complementary factor parameter, so as to make up for the decrease of cleaning efficiency caused by the decrease of the cleaning comprehensive index, and realize the dynamic balance of the knowledge expression layer to the cleaning efficiency and the recontamination risk.

[0078] The preset complementary factor parameter is the cleaning oil temperature or the temperature of the spraying medium. By increasing the cleaning oil temperature or the temperature of the spraying medium, the viscosity of the oil is reduced and the sediment stripping capacity is enhanced, so that the cleaning effect is stable under the condition of reducing the cleaning comprehensive index.

[0079] In the embodiment of the present application, the purpose of step S400 is to judge the local oil flow disturbance intensity of the current to-be-cleaned region based on the recontamination intensity threshold value determined based on the historical cleaning task data, and to directly set the cleaning comprehensive index corresponding to the node as the reference cleaning comprehensive index adopted this time when it is determined that the local oil flow disturbance intensity exceeds the recontamination intensity threshold value. The setting is based on the following understanding: for the combination of the structural characteristics and the local oil flow disturbance intensity of the current to-be-cleaned region, the historical data has shown that the recontamination index will continue to exceed the preset threshold value after exceeding the node, and continuing to improve the cleaning comprehensive index will lead to uncontrollable recontamination. Therefore, limiting the cleaning comprehensive index at the level corresponding to the node can achieve the maximum cleaning comprehensive index within the acceptable range without causing uncontrollable recontamination, in other words, the cleaning comprehensive index is controlled to be in a critical but controllable interval.

[0080] In the embodiment of the present application, the purpose of step S500 is to increase the preset complementary factor parameter on the basis of limiting the cleaning comprehensive index in step S400, so that the overall cleaning effect is compensated and stabilized after the increase and the decrease. Specifically, after the cleaning comprehensive index is limited at the level corresponding to the recontamination intensity threshold value, the preset complementary factor parameter, i.e., the cleaning oil temperature or the injection medium temperature, is increased to reduce the oil viscosity and enhance the stripping capacity, so as to compensate and stabilize the cleaning effect on the basis of the maximum and acceptable cleaning comprehensive index.

[0081] In the embodiment of the present application, step S501 uses the "increase amplitude of the current local oil flow disturbance intensity compared to the recontamination intensity threshold value" as the correction factor, because the increase amplitude can quantitatively reflect the exceeding degree of the current environment relative to the critical state. The increase amplitude can be an absolute difference value between the two or a deviation ratio of the two. When the difference value form is used, the excess amount of the current disturbance environment relative to the threshold value can be directly reflected, which is suitable for the case where the dimensions are consistent and the monitoring accuracy is high; when the deviation ratio form is used, the relative change rate of the disturbance intensity relative to the critical value can be reflected, which is more suitable for the normalized comparison under different devices and different oil flow conditions, thereby enhancing the adaptability and cross-device universality of the correction factor.

[0082] From the perspective of knowledge expression, when the cleaning comprehensive index is limited at the critical point level, the dominant variable leading to uncontrollable recontamination is transformed into "environment disturbance excess amount", which directly determines the amplitude that needs to be compensated by the preset complementary factor parameter. Using the increase amplitude as the correction factor has two advantages: one is that the dimension is directly related to the risk boundary, avoiding the introduction of redundant variables that have no direct correspondence with recontamination; the other is that comparability can be achieved through normalization between different devices and different conditions, thereby ensuring the portability and repeatability of the adjustment strategy.

[0083] In the embodiment of the present application, the preset correction amplitude coefficient in step S502 is used to map the correction factor to the actual adjustment amount of the preset complementary factor parameter, which can be set according to prior art means, such as fitting through historical cleaning task data, calibration test or empirical rules, including empirical relationships based on oil viscosity-temperature curve, material surface energy-temperature response curve, and cleaning effect-temperature rise amplitude. The coefficient can be set as a fixed value, or a segmented or adaptive value according to the structural features and fouling conditions of the area to be cleaned. The method of "acting on the preset complementary factor parameter" in step S502 can be direct multiplication amplification, or additive superposition, exponential mapping or segmented linear function, etc. The specific selection can be determined by the implementation system according to control stability and energy consumption constraints.

[0084] In the embodiment of the present application, the following examples are used to illustrate the complete process from structural feature identification, sample screening to preset complementary factor parameter revision. Assuming that the structural features of a certain area to be cleaned are: channel width 4.0 mm, channel depth 18.0 mm, surface roughness Ra=2.5 μm, and fouling thickness 0.35 mm, a number of sample data consistent with the above structural features but different in cleaning comprehensive index are screened out from the historical cleaning task data. After sorting and trend analysis of these sample data, it is found that the recontamination index changes slowly in the interval where the cleaning comprehensive index M increases from 0.60 to 0.72, the recontamination index first exceeds the preset threshold when M=0.75, and the index continues to exceed the preset threshold when M>0.75, and the corresponding local oil flow disturbance intensity is =0.40 (normalized dimension). Therefore, the recontamination intensity threshold is determined to be =0.40, and the cleaning comprehensive index of the node is M*=0.75. The real-time measured local oil flow disturbance intensity of the current task is R=0.46, which is higher than the recontamination intensity threshold . Therefore, the cleaning comprehensive index of this time is set to M=M*=0.75 in step S400, and is not further improved. Then, the correction factor is calculated in step S501, assuming that , and ΔR is taken as the correction factor. In step S502, assuming that the preset correction amplitude coefficient k=80 ℃ (after unitization, it represents the oil temperature adjustment amplitude corresponding to each unit of disturbance excess), the oil temperature adjustment amount = . If the current cleaning oil temperature is 55.0 ℃, the adjusted cleaning oil temperature is about 59.8 ℃. Under this adjustment strategy, the cleaning comprehensive index is limited to the maximum acceptable value M* that will not cause recontamination out of control, and the stripping ability is compensated through the gain of the preset complementary factor parameter, so that the cleaning effect reaches the expectation. If the injection medium temperature is taken as the preset complementary factor parameter, the injection medium temperature can be revised in the same way.

[0085] In this embodiment of the invention, the overall beneficial effects are manifested as follows: At the knowledge representation level, a calculable critical correlation is established between the comprehensive cleaning index and the risk of recontamination. The comprehensive cleaning index is limited to a controllable boundary by a recontamination intensity threshold, avoiding an increase in the migration, attachment, or deposition of pollutants in non-target areas due to excessively high comprehensive cleaning indices. After limiting the comprehensive cleaning index, quantitative compensation through preset complementary factor parameters ensures the overall cleaning effect remains stable, thereby achieving a quantifiable dynamic balance between cleaning efficiency and recontamination risk. This effect directly corresponds to the core research point described in step S200: for current areas to be cleaned with the same structural characteristics but different local oil flow disturbance intensities, how to achieve reasonable control of recontamination and reasonable improvement related to oil temperature at the knowledge representation and understanding level. This invention provides a clear parameterized solution through a compensation mechanism using the improvement magnitude as a correction factor.

[0086] In this embodiment of the invention, the application prospects are reflected in the internal cleaning operations of various types of oil-immersed power transformers, which can achieve adaptive strategy generation under different structural characteristics and different local oil flow disturbance intensities; during long-term operation maintenance cycles, the recontamination intensity threshold and preset correction amplitude coefficient can be optimized by continuously accumulating historical cleaning task data, thereby improving the system's generalization ability and energy efficiency under complex operating conditions; in energy-constrained wireless operation scenarios, by limiting the comprehensive cleaning index and using preset complementary factor parameters for compensation, it is beneficial to save energy consumption and improve the quality and reliability of task completion under the premise of controllable recontamination.

[0087] Example 2:

[0088] Figure 4 An application architecture diagram of the system provided in an embodiment of the present invention is shown.

[0089] In another preferred embodiment of the present invention, a robot knowledge representation system based on structure recognition includes:

[0090] The structure recognition module 100 is used to determine the structural features of the area to be cleaned inside the power transformer that the cleaning robot is dealing with, and to acquire historical cleaning task data of the cleaning robot.

[0091] The structural features of the area to be cleaned include the geometric morphological parameters, channel spatial distribution characteristics, surface roughness parameters, and scaling conditions of the area to be cleaned.

[0092] Furthermore, the structure recognition-based robot knowledge representation system also includes:

[0093] The sample analysis module 200 is used to analyze historical cleaning task data and obtain several sample data of the cleaning robot when it faces a structural feature that is consistent with the current area to be cleaned but has different specific comprehensive cleaning indicators.

[0094] The specific cleaning comprehensive index refers to a composite parameter representing the cleaning effect intensity applied by the cleaning robot during the cleaning process, including at least one of the jet flow rate or the jet pressure pulse frequency or a combination of both.

[0095] Further, the robot knowledge expression system based on structure recognition further comprises:

[0096] The recontamination identification module 300 is configured to obtain the recontamination index corresponding to each sample data after cleaning, and analyze the node at which the recontamination index first exceeds the preset threshold when the cleaning comprehensive index gradually increases, and determine the local oil flow disturbance intensity corresponding to the node.

[0097] Further, the robot knowledge expression system based on structure recognition further comprises:

[0098] The disturbance analysis module 400 is configured to obtain the local oil flow disturbance intensity of the current area to be cleaned, and analyze whether it exceeds the local oil flow disturbance intensity corresponding to the node; if so, the cleaning comprehensive index corresponding to the node is set as the reference cleaning comprehensive index adopted in this cleaning.

[0099] Further, the robot knowledge expression system based on structure recognition further comprises:

[0100] The knowledge correction module 500 is configured to generate a correction factor according to the difference between the current local oil flow disturbance intensity and the local oil flow disturbance intensity corresponding to the node, and correct the preset complementary factor parameter according to the correction factor.

[0101] It should be understood that, although each step in the flowchart of each embodiment of the present application is displayed in sequence according to the arrow, these steps are not necessarily executed in the order indicated by the arrow. Unless otherwise specified herein, the execution of these steps is not strictly limited in sequence, and these steps can be executed in other orders. Moreover, at least part of the steps in each embodiment can include multiple sub-steps or multiple stages, which are not necessarily executed at the same time, but can be executed at different times, and the execution order of these sub-steps or stages is not necessarily sequential, but can be executed in rotation or alternation with at least part of other steps or sub-steps or stages of other steps.

[0102] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by instructing the relevant hardware through a computer program. The program can be stored in a non-volatile computer readable storage medium, and when the program is executed, the processes of the above-mentioned embodiments of the methods can be included. Any reference to memory, storage, database, or other medium used in the embodiments provided in the present application can include non-volatile and / or volatile memory. Non-volatile memory can include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), or flash memory. Volatile memory can include random access memory (RAM) or external cache memory. As an illustration but not limitation, RAM is available in various forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDR SDRAM), enhanced SDRAM (ESDRAM), synchronous link (SynchlMnk) DRAM (SLDRAM), memory bus (Rambus) direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM), etc.

[0103] Any combination of the technical features of the above-mentioned embodiments can be combined. In order to make the description simple, all possible combinations of the technical features in the above-mentioned embodiments are not described, however, as long as the combination of the technical features does not exist contradictory, it should be considered as the scope of the present application.

[0104] The above-mentioned embodiments only express several embodiments of the present application, and the description is more specific and detailed, but it should not be understood as limiting the scope of the present application. It should be noted that for ordinary skilled in the art, without departing from the concept of the present application, a number of modifications and improvements can be made, which are within the scope of the present application. Therefore, the protection scope of the present application should be subject to the appended claims.

[0105] The above-mentioned embodiments are only the preferred embodiments of the present application, and are not used to limit the present application. Any modification, equivalent replacement, and improvement made within the spirit and principle of the present application should be included in the protection scope of the present application.

Claims

1. A method for representing knowledge of a robot based on structure recognition, characterized by, The method comprises: determining the structural characteristics of the internal area to be cleaned of the power transformer that the cleaning robot should face, and obtaining historical cleaning task data of the cleaning robot; analyzing the historical cleaning task data to obtain a plurality of sample data that are consistent with the structural characteristics of the current area to be cleaned but have different cleaning comprehensive indexes; the cleaning comprehensive index refers to a composite parameter representing the cleaning effect intensity of the cleaning robot in the cleaning process, including at least one of jet flow rate or jet pressure pulse frequency; obtaining the post-cleaning recontamination index corresponding to each sample data, and analyzing the node at which the recontamination index first exceeds the preset threshold when the cleaning comprehensive index gradually increases, and determining the local oil flow disturbance intensity corresponding to the node; obtaining the local oil flow disturbance intensity of the current area to be cleaned, and analyzing whether it exceeds the local oil flow disturbance intensity corresponding to the node; if so, the cleaning comprehensive index corresponding to the node is set as the reference cleaning comprehensive index adopted in this cleaning; generating a correction factor according to the difference between the current local oil flow disturbance intensity and the local oil flow disturbance intensity corresponding to the node, and correcting the preset complementary factor parameter according to the correction factor; the preset complementary factor parameter is the cleaning oil temperature or the jet medium temperature. 2.The structure recognition based robot knowledge representation method according to claim 1, characterized in that, The structural characteristics of the area to be cleaned include the geometric shape parameters, channel space distribution characteristics, surface roughness parameters and fouling conditions of the area to be cleaned. 3.The structure recognition based robot knowledge representation method according to claim 1, characterized in that, The cleaning comprehensive indexes adopted by the cleaning robot when facing different sample data consistent with the structural characteristics of the current area to be cleaned are different, and the reason for the difference is that the local oil flow disturbance intensities of the cleaned areas corresponding to the sample data are different. 4.The structure recognition based robot knowledge representation method according to claim 1, characterized in that, The steps of obtaining the post-cleaning recontamination index corresponding to each sample data, and analyzing the node at which the recontamination index first exceeds the preset threshold when the cleaning comprehensive index gradually increases, and determining the local oil flow disturbance intensity corresponding to the node, comprise: sequentially analyzing each sample data to obtain the recontamination index of the cleaning robot after completing the cleaning of the internal power transformer, the recontamination index being used to represent the degree of contamination of non-target areas caused by the increase of jet flow rate or jet pressure pulse frequency under the action of the cleaning comprehensive index; sorting the different sample data according to the cleaning comprehensive index from small to large, and analyzing the change trend of the recontamination index under the sorting; if there is a node at which the recontamination index first exceeds the preset threshold in the change trend of the recontamination index, and the recontamination indexes after the node all exceed the preset threshold, then the local oil flow disturbance intensity corresponding to the node is determined as the recontamination intensity threshold.

5. The structure recognition based robot knowledge representation method according to claim 4, characterized in that, The steps of generating a correction factor according to the difference between the current local oil flow disturbance intensity and the local oil flow disturbance intensity corresponding to the node, and correcting the preset complementary factor parameter according to the correction factor, comprise: calculating the improvement amplitude of the current local oil flow disturbance intensity compared with the recontamination intensity threshold, and taking the improvement amplitude as the correction factor; multiplying the correction factor and the preset correction amplitude coefficient to jointly act on the preset complementary factor parameter.

6. A structure recognition based robot knowledge representation system, characterized by, The system comprises: The structure recognition module is configured to determine a structural feature of an internal area to be cleaned of a power transformer that the cleaning robot is to deal with, and to obtain historical cleaning task data of the cleaning robot. The sample analysis module is configured to analyze the historical cleaning task data, and to obtain a plurality of sample data that are consistent with the structural feature of the current area to be cleaned but have different cleaning comprehensive indexes. The cleaning comprehensive index refers to a composite parameter representing a cleaning action intensity applied by the cleaning robot during cleaning, and includes at least one of a jet flow rate or a jet pressure pulse frequency. The recontamination identification module is configured to obtain a recontamination index corresponding to each sample data after cleaning, and to analyze a node at which the recontamination index first exceeds a preset threshold when the cleaning comprehensive index gradually increases, and to determine a local oil flow disturbance intensity corresponding to the node. The disturbance analysis module is configured to obtain a local oil flow disturbance intensity of the current area to be cleaned, and to analyze whether the local oil flow disturbance intensity exceeds the local oil flow disturbance intensity corresponding to the node. If so, the cleaning comprehensive index corresponding to the node is set as a reference cleaning comprehensive index to be used in the current cleaning. The knowledge correction module is configured to generate a correction factor according to a difference between the current local oil flow disturbance intensity and the local oil flow disturbance intensity corresponding to the node, and to correct a preset complementary factor parameter according to the correction factor. The preset complementary factor parameter is a cleaning oil temperature or a jet medium temperature.

7. The structure recognition based robot knowledge representation system of claim 6, wherein, The structural feature of the area to be cleaned includes a geometric parameter, a channel space distribution feature, a surface roughness parameter, and a fouling condition of the area to be cleaned.

Images