Intelligent energy multi-energy evaluation visualization system and method for industry

Abstract

The application discloses a kind of wisdom energy multi-energy evaluation visualization system and method for industry, it is related to energy big data visualization cross technical field, including, acquisition multi-source heterogeneous data and pre-processing, by space-time hash coding method, generate high-dimensional energy data of space-time alignment;Based on three-dimensional manifold coordinate model, the equipment energy efficiency parameter is converted into equivalent energy quality field, by energy field reconstruction and geometric optimization method, generate curved manifold space and identify optimal transmission path and distortion intensity cloud chart;Based on distortion intensity cloud chart, generate thermal color scale by thermal color scale mapping algorithm, and the three-dimensional gradient field distribution of equivalent energy quality field is rendered using ray tracing method, generate visual energy evaluation interface.The application generates space and time hash code by space-time hash coding method, establishes equipment-grid mapping relationship in combination with GeoHash coding, realizes the accurate space-time alignment of multi-source heterogeneous data.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of energy big data visualization intersection, and particularly relates to a smart energy multi-energy evaluation visualization system and method for industry. BACKGROUND

[0002] In the field of industrial energy management, the evaluation and visualization of multi-energy systems are key links for optimizing energy efficiency. Existing technologies usually adopt time series-based monitoring methods to collect parameters such as voltage, temperature, and pressure through sensors, and evaluate energy efficiency status using statistical analysis or machine learning models. For example, some systems integrate multi-source heterogeneous data through multi-dimensional data fusion technology, use principal component analysis (PCA) or t-SNE algorithm for dimension reduction, and generate two-dimensional or three-dimensional visualization results. In addition, some methods combine threshold division of energy efficiency indicators (such as power factor and entropy increase rate) to mark the regions of equipment operation status, assisting energy management personnel to identify abnormal regions. These methods can reflect the operation characteristics of energy systems to some extent, providing data support for decision-making.

[0003] However, existing methods have limitations in time-space alignment and high-dimensional data expression. The differences in time-space resolution of multi-source heterogeneous data may lead to distortion of the fusion results, affecting the accuracy of evaluation. For example, the sampling frequency and spatial distribution of voltage harmonic distortion rate and pipeline temperature gradient are different, and direct fusion may introduce time-space deviation. In addition, traditional dimension reduction methods are difficult to simultaneously preserve local neighborhood structure and global topological relationship, leading to incomplete geometric expression of energy efficiency characteristics. SUMMARY

[0004] In view of the above existing problems, the present application is proposed.

[0005] Therefore, the present application provides a smart energy multi-energy evaluation visualization method for industry to solve the problems of insufficient time-space alignment accuracy of multi-source heterogeneous data and incomplete high-dimensional energy efficiency feature manifold expression.

[0006] To solve the above technical problems, the present application provides the following technical solutions:

[0007] In a first aspect, the present application provides a method for visualizing multi-energy evaluation of smart energy for industry, which comprises collecting and preprocessing multi-source heterogeneous data, generating high-dimensional energy data aligned in space-time by a space-time hashing coding method, the multi-source heterogeneous data including voltage harmonic distortion rate, pipeline temperature gradient, storage tank pressure fluctuation, power factor, entropy increase rate and frequency fluctuation, performing local neighborhood optimization and global topology preservation by a UMAP manifold learning algorithm, mapping the high-dimensional energy data aligned in space-time into a three-dimensional manifold coordinate model with energy efficiency labels, converting equipment energy efficiency parameters into equivalent energy quality field based on the three-dimensional manifold coordinate model, generating curved manifold space and identifying optimal transmission path and distortion intensity cloud map by energy field reconstruction and geometric optimization method, generating thermal color scale by a thermal color scale mapping algorithm based on the distortion intensity cloud map, and rendering three-dimensional gradient field distribution of the equivalent energy quality field by ray tracing method to generate a visual energy evaluation interface.

[0008] In a preferred embodiment of the method for visualizing multi-energy evaluation of smart energy for industry, the preprocessing comprises denoising, outlier rejection and sliding window smoothing of the multi-source heterogeneous data.

[0009] The high-dimensional energy data aligned in space-time is generated by the space-time hashing coding method as follows,

[0010] The industrial plant physical space is divided into a three-dimensional grid space by a spatial discretization algorithm.

[0011] A space hash code is generated by an MD5 hashing algorithm, and a time stamp hash code is generated by a time window and a CRC32 algorithm.

[0012] A device-grid mapping relationship is established by a GeoHash coding algorithm, and the space hash code and the time stamp hash code are aligned in space-time according to the device-grid mapping relationship to generate the high-dimensional energy data aligned in space-time.

[0013] In a preferred embodiment of the method for visualizing multi-energy evaluation of smart energy for industry, the local neighborhood optimization and global topology preservation by the UMAP manifold learning algorithm are performed as follows,

[0014] The Euclidean distance matrix of the high-dimensional energy data is obtained by a Minkowski distance measurement method, and the Euclidean distance matrix is converted into a probability distribution map using an exponential kernel function.

[0015] An initial coordinate point set is randomly generated in the three-dimensional grid space by a uniform distribution random sampling method.

[0016] The initial coordinate point set is taken as the starting point of the optimization process, and the coordinates of the three-dimensional grid cells are iteratively optimized by minimizing the cross-entropy loss between the high-dimensional probability graph and the low-dimensional embedding through the stochastic gradient descent algorithm, to generate three-dimensional manifold coordinates with global topology preservation characteristics.

[0017] As a preferred scheme of the intelligent energy multi-energy evaluation visualization method for industry, wherein: the mapping of the spatio-temporal aligned high-dimensional energy data into a three-dimensional manifold coordinate model with energy efficiency labels refers to defining an entropy increase rate threshold, dividing the energy efficiency region into high-efficiency and low-efficiency zones by SVM based on the three-dimensional manifold coordinates, combining power factor and entropy increase rate threshold, and assigning energy efficiency labels to generate a manifold coordinate model containing three-dimensional coordinates and energy efficiency labels.

[0018] As a preferred scheme of the intelligent energy multi-energy evaluation visualization method for industry, wherein: the conversion of the device energy efficiency parameters into an equivalent energy quality field refers to mapping the device energy efficiency parameters in the three-dimensional manifold coordinate model into an energy scalar field by the Gaussian kernel density estimation method and assigning values to the energy efficiency region, and mapping the power factor field and the entropy increase rate change into the equivalent energy quality field based on the energy scalar field through convolution integration.

[0019] As a preferred scheme of the intelligent energy multi-energy evaluation visualization method for industry, wherein: the generation of a curved manifold space and the identification of optimal transmission paths and distortion intensity cloud maps through the energy field reconstruction and geometric optimization method are as follows,

[0020] The equivalent energy quality field is used to construct the Riemann metric tensor and establish the geometric structure of the energy space.

[0021] The connection coefficients are obtained through the Christoffel symbol, and the curved manifold space is generated by combining the Riemann curvature tensor.

[0022] The initial flow direction of the equivalent energy quality field in the curved manifold space is defined, and the Eikonal equation is solved by the fast marching method to identify the optimal transmission path of the equivalent energy quality field in the curved manifold space.

[0023] The distortion intensity of the optimal transmission path is calculated by using the Riemann geometric curvature and gradient coupling algorithm.

[0024] The distortion intensity threshold is defined and compared with the distortion intensity of the optimal transmission path to identify the energy black hole region.

[0025] Based on the identified energy black hole region, the Marching Cubes algorithm is used to generate a distortion intensity cloud map.

[0026] As a preferred scheme of the intelligent energy multi-energy evaluation visualization method for industry provided by the present application, wherein: the thermal color scale is generated by a thermal color scale mapping algorithm, and a three-dimensional gradient field distribution of the equivalent energy quality field is rendered by a ray tracing method to generate a visual energy evaluation interface, and the steps are as follows,

[0027] A continuous color scale ruler from cold to warm is generated by a thermal color scale mapping algorithm based on the distortion intensity cloud map;

[0028] The equivalent energy quality field is colored by a color scale-based optical property conversion method;

[0029] The colored equivalent energy quality field is taken as a non-uniform participating medium, and a pseudo-color volume rendering image is generated by a Monte Carlo path tracing algorithm;

[0030] The pseudo-color volume rendering image, the energy efficiency black hole isosurface, and the three-dimensional grid space are fused by FBO to generate a visual energy evaluation interface.

[0031] In a second aspect, the present application provides an intelligent energy multi-energy evaluation visualization system for industry, comprising a data acquisition module, a three-dimensional manifold generation module, a cloud map generation module, and a visualization module; the data acquisition module is used for acquiring multi-source heterogeneous data and preprocessing, generating high-dimensional energy data aligned in time and space by a time-space hash coding method; the three-dimensional manifold generation module is used for local neighborhood optimization and global topology preservation by a UMAP manifold learning algorithm, and mapping the high-dimensional energy data aligned in time and space into a three-dimensional manifold coordinate model with energy efficiency labels; the cloud map generation module is used for converting device energy efficiency parameters into an equivalent energy quality field based on the three-dimensional manifold coordinate model, generating a curved manifold space and identifying an optimal transmission path and a distortion intensity cloud map by an energy field reconstruction and geometric optimization method; and the visualization module is used for generating a thermal color scale by a thermal color scale mapping algorithm based on the distortion intensity cloud map, and rendering a three-dimensional gradient field distribution of the equivalent energy quality field by a ray tracing method to generate a visual energy evaluation interface.

[0032] In a third aspect, the present application provides a computer device comprising a memory and a processor, wherein the memory stores a computer program, and wherein the computer program is executed by the processor to implement any step of the intelligent energy multi-energy evaluation visualization method for industry according to the first aspect of the present application.

[0033] In a fourth aspect, the present application provides a computer readable storage medium having a computer program stored thereon, wherein the computer program is executed by a processor to implement any step of the intelligent energy multi-energy evaluation visualization method for industry according to the first aspect of the present application.

[0034] The application has the beneficial effects that: a space-time hash code is generated by a space-time hash coding method, a device-grid mapping relationship is established in combination with a GeoHash coding, accurate space-time alignment of multi-source heterogeneous data is realized, and a high-precision data basis is provided for subsequent analysis; and through Minkowski distance measurement and random gradient descent optimization, high-dimensional data is mapped into three-dimensional manifold coordinates that maintain global topological structure, and the essential distribution characteristics of energy data are revealed. The two steps together realize a complete analysis chain from data preprocessing to space mapping, enable the energy efficiency evaluation result to be more accurate and reliable, and provide key technical support for intelligent management of industrial energy systems. BRIEF DESCRIPTION OF DRAWINGS

[0035] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings needed in the embodiment description will be briefly introduced. Obviously, the drawings in the following description are only some embodiments of the present application, and other drawings can be obtained by those skilled in the art without creative labor.

[0036] Fig. 1 Flowchart of the intelligent energy multi-energy evaluation visualization method for industry.

[0037] Fig. 2 Schematic diagram of the intelligent energy multi-energy evaluation visualization system for industry.

[0038] Fig. 3 Flowchart of the space-time hash coding method.

[0039] Fig. 4 Flowchart of the energy field reconstruction and geometric optimization. DETAILED DESCRIPTION

[0040] In order to make the above-mentioned purposes, features and advantages of the present application more apparent and easy to understand, the specific embodiments of the present application will be described in detail below with reference to the drawings of the specification.

[0041] In the following description, many specific details are set forth in order to provide a thorough understanding of the present application, but the present application can also be implemented in other ways different from those described herein, and those skilled in the art can make similar generalizations without departing from the connotation of the present application, therefore the present application is not limited to the specific embodiments disclosed below.

[0042] Secondly, the "one embodiment" or "embodiment" referred to herein means that the specific features, structures or characteristics can be included in at least one implementation of the present application. "In one embodiment" appearing in different places in the specification does not mean the same embodiment, nor is it an independent or alternative embodiment that excludes other embodiments.

[0043] ReferenceFigs. 1-4 For an embodiment of the present application, the embodiment provides a smart energy multi-energy evaluation visualization method for industry, comprising the following steps:

[0044] S1, collect multi-source heterogeneous data and preprocess, generate high-dimensional energy data aligned in time and space through space-time hash coding method;

[0045] The multi-source heterogeneous data includes voltage harmonic distortion rate, pipeline temperature gradient, storage tank pressure fluctuation, power factor, entropy increase rate and frequency fluctuation;

[0046] Further, the voltage harmonic distortion rate is obtained by real-time monitoring of the alternating voltage waveform by the power quality sensor and calculating the percentage of each harmonic content. The pipeline temperature gradient is measured by the distributed optical fiber temperature sensor along the axial direction of the pipeline, and the temperature difference between adjacent temperature measuring points is measured. The storage tank pressure fluctuation is recorded by the hydrogen energy pressure transmitter with millisecond level sampling frequency. The power factor is calculated by synchronously measuring the phase difference between voltage and current by three-phase electric energy meter. The entropy increase rate is calculated by Clausius integral formula according to the heat flux density and temperature gradient data collected by the thermodynamic sensor. The frequency fluctuation is obtained by continuously monitoring the power frequency period change by high-precision frequency meter.

[0047] The preprocessing includes denoising, outlier rejection and sliding window smoothing processing of multi-source heterogeneous data;

[0048] Further, the voltage harmonic distortion rate data is processed by wavelet threshold denoising method, and the high-frequency coefficients are filtered by soft threshold after multi-scale decomposition to eliminate noise. The pipeline temperature gradient data is detected by Grubbs criterion, and the measured value is removed when the deviation from the sliding average value exceeds three times the standard deviation. The storage tank pressure fluctuation data is smoothed by Hann window weighted sliding window, and the window width is self-adaptive to the pressure change rate. The power factor data is removed by Kalman filter algorithm to keep the integrity of the phase characteristics. The entropy increase rate data is processed by median filter combined with linear interpolation method. The frequency fluctuation data uses Butterworth low-pass filter to suppress high-frequency interference.

[0049] The industrial plant physical space is divided into three-dimensional grid space by space discretization algorithm, and each three-dimensional grid unit is assigned a unique three-dimensional coordinate;

[0050] Further, when the space discretization algorithm processes the industrial plant physical space, first determine the resolution parameter of space division, divide the length, width and height directions of the plant according to the same interval, form a regular array of cubic units. Each cubic unit obtains a unique coordinate identifier by combining its X-axis, Y-axis and Z-axis direction serial number, and the coordinate value records the spatial position of the unit center point in the form of floating point number.

[0051] Based on the three-dimensional grid space and the device type, a space hash code is generated by an MD5 hash algorithm, and a time stamp hash code is generated by a time window and a CRC32 algorithm.

[0052] Further, when the MD5 hash algorithm processes the coordinates of the three-dimensional grid unit and the device type, the floating-point coordinate value is first converted into a fixed format string, which is spliced with the device type string. After the input string is processed by padding, it is divided into 512-bit groups. Each group is processed by four rounds of nonlinear function, each round containing 16 bit operations. Finally, the space hash code is output. When the time window processes the timestamp data, the start and end time of the time window is first determined, and the year, month, day, hour, minute, second and millisecond information of all time points in the window is spliced into a string. The CRC32 algorithm performs polynomial division operation on the string to generate the timestamp hash code through 32 times of shift and XOR operation.

[0053] It should be noted that the device type specifically refers to the device classification identification in the industrial plant, including: generator, transformer, heat exchanger, hydrogen storage tank and frequency converter, etc.

[0054] By using the GeoHash encoding algorithm, a device-grid mapping relationship is established, and according to the device-grid mapping relationship, the space-time alignment of the space hash code and the time stamp hash code is performed to generate the space-time aligned high-dimensional energy data.

[0055] Further, the GeoHash encoding algorithm converts the device latitude, longitude and elevation coordinates into a string encoding. The length of the string determines the spatial accuracy. The first few characters represent a larger grid area, and the last few characters represent a finer sub-grid. The device type string is spliced with the GeoHash code as the input of the space hash code. The timestamp hash code is generated by processing the year, month, day, hour, minute, second and millisecond string through the CRC32 algorithm. The space-time alignment process first matches the device data with the same space hash code to ensure that the data comes from the same grid unit. Then, the time stamp hash codes are compared, and the data with a time difference less than the sampling interval is considered to be synchronized. Finally, the voltage harmonic distortion rate, pipeline temperature gradient, storage tank pressure fluctuation, power factor, entropy increase rate and frequency fluctuation are organized in time sequence to form a two-dimensional data matrix with rows representing time points and columns representing parameter types. Each data unit in the two-dimensional data matrix is associated with a corresponding space hash code and time stamp hash code. The rows of the two-dimensional data matrix are sorted by the time stamp hash code, and the columns are arranged in a fixed order according to the parameter type. The space-time alignment of multi-source heterogeneous data is realized, and the space-time aligned high-dimensional energy data is finally generated.

[0056] S2, local neighborhood optimization and global topology preservation are performed by UMAP manifold learning algorithm, and the space-time aligned high-dimensional energy data is mapped into a three-dimensional manifold coordinate model with energy efficiency label;

[0057] Based on the spatio-temporal alignment of high-dimensional energy data, the Euclidean distance matrix of high-dimensional energy data is obtained by Minkowski distance measurement method, and the Euclidean distance matrix is converted into a probability distribution map using an exponential kernel function;

[0058] Further, the high-dimensional energy data is first subjected to spatio-temporal alignment preprocessing to ensure that the timestamps and spatial coordinates of all data points are strictly matched. The Minkowski distance measurement method is used to process the aligned high-dimensional energy data. This method can be degenerated into the standard Euclidean distance formula through parameter adjustment. For each pair of data points, the sum of the squares of the dimensional value differences is squared according to the definition of Euclidean distance to obtain the distance value between the two points. The distance values between all pairs of data points are filled into the Euclidean distance matrix to form a complete Euclidean distance matrix. The Euclidean distance matrix reflects the distribution relationship of high-dimensional energy data in the feature space. The Euclidean distance matrix is then converted and processed by an exponential kernel function. The exponential kernel function takes the natural constant e as the base number and takes each distance value in the Euclidean distance matrix as the input of the negative exponential function. After exponential transformation, the original distance value is mapped to a probability value between 0 and 1. The smaller the distance, the closer the probability value to 1, and the larger the distance, the probability value tends to 0. The converted result is a probability distribution map, and each element in the probability distribution map represents the similarity probability between the corresponding data points.

[0059] An initial coordinate point set is randomly generated in a three-dimensional grid space by a uniform distribution random sampling method;

[0060] Further, an initial coordinate point set is generated in the effective range of the three-dimensional grid space by a uniform distribution random sampling method. The uniform distribution random sampling method ensures that each coordinate point is independently and randomly distributed in the x, y, and z dimensions, and all points are strictly located within the grid boundary. The generated initial coordinate point set is uniformly distributed in space without clustering, and each point has an equal probability of occurrence. The final output is an initial coordinate point set that meets the uniformity requirement.

[0061] The initial coordinate point set is used as the starting point of the optimization process, and the coordinates of the three-dimensional grid cells are iteratively optimized by the stochastic gradient descent algorithm to minimize the cross-entropy loss between the high-dimensional probability map and the low-dimensional embedding, generating a three-dimensional manifold coordinate with global topology preservation characteristics;

[0062] Further, the initial set of coordinate points serves as the starting point for the optimization process, and the coordinates of the three-dimensional grid cells are iteratively optimized using the stochastic gradient descent algorithm. During the optimization process, the cross-entropy loss function value between the current three-dimensional grid cell coordinates and the high-dimensional probability map is first calculated. This loss function quantifies the difference in probability distribution between the low-dimensional embedding and the high-dimensional data. The stochastic gradient descent algorithm randomly selects a portion of data points at each iteration, calculates the loss function gradient for these points, and updates the coordinates of the three-dimensional grid cells based on the gradient direction. The adaptive learning rate strategy is used for coordinate updating to ensure that the optimization process maintains sufficient convergence speed while avoiding oscillation. As the iteration progresses, the spatial structure of the low-dimensional embedding is gradually adjusted, so that the distance between similar data points on the three-dimensional manifold reflects their similarity in the high-dimensional probability map. The optimization process continues until the cross-entropy loss changes by less than the convergence threshold (with a value range of 1e-5 to 1e-3) for multiple consecutive iterations, indicating that the algorithm has reached a stable convergence state. The final generated three-dimensional manifold coordinates completely preserve the global topological relationship of the high-dimensional data, with similar data points remaining adjacent and dissimilar data points being far apart on the manifold.

[0063] Based on statistical analysis of historical operation data of industrial plant equipment, an entropy increase rate threshold is defined.

[0064] Further, based on statistical analysis of historical operation data of industrial plant equipment, the definition process of the entropy increase rate threshold is as follows: first, collect the historical operation parameters of the industrial plant equipment under normal and abnormal operating conditions, including time series data of key indicators such as temperature, vibration, and energy consumption; then calculate the entropy increase rate of the historical operation data of the industrial plant equipment in each time window; through kernel density estimation, statistically analyze the entropy increase rate distribution of a large number of normal and abnormal samples to determine the critical value that can effectively distinguish between the two states; finally, select the entropy value with the highest classification accuracy as the entropy increase rate threshold.

[0065] According to the three-dimensional manifold coordinates, combined with the power factor and the entropy increase rate threshold, the energy efficiency area is divided into high-efficiency and low-efficiency regions using SVM, and energy efficiency labels are assigned, generating a manifold coordinate model containing three-dimensional coordinates and energy efficiency labels.

[0066] Further, based on the three-dimensional manifold coordinates, power factor and entropy rate threshold, the specific process of energy efficiency region division using support vector machine (SVM) is as follows: first, the three-dimensional manifold coordinates are taken as spatial features, the power factor and the entropy rate are taken as energy efficiency features, and the features are collectively input into the SVM classifier to form a feature vector. The SVM uses a radial basis kernel function to construct an optimal classification hyperplane in a high-dimensional feature space, and divides the data into two categories of high-efficiency region and low-efficiency region by maximizing the classification interval. In the training process, the cross-validation method is used to determine the penalty parameter and the kernel function parameter of the SVM to ensure that the classifier has the best generalization performance. The trained SVM classifier predicts each coordinate point on the three-dimensional manifold, and according to the comparison result of the power factor and the entropy rate with the threshold, an energy efficiency label of "high efficiency" or "low efficiency" is assigned to each region. The finally generated three-dimensional manifold coordinate model contains three-dimensional space coordinates and corresponding energy efficiency labels, which completely characterizes the energy efficiency distribution characteristics of the industrial plant equipment in the three-dimensional manifold space.

[0067] It should be noted that the three-dimensional manifold space is a continuous energy field distribution space formed by Gaussian kernel density estimation of energy efficiency features based on the three-dimensional manifold coordinate model. In the three-dimensional manifold space, the position and attributes (such as energy efficiency features) of each coordinate point are determined by the corresponding data points in the three-dimensional manifold coordinate model.

[0068] S3, based on the three-dimensional manifold coordinate model, the device energy efficiency parameters are converted into equivalent energy quality field, and the curved manifold space is generated and the optimal transmission path and the distortion intensity cloud map are identified by energy field reconstruction and geometric optimization method;

[0069] By the Gaussian kernel density estimation method, the device energy efficiency parameters in the three-dimensional manifold coordinate model are mapped into an energy scalar field and the energy efficiency region is valued, wherein the high-efficiency region is assigned a positive value and the low-efficiency region is assigned a negative value.

[0070] Further, based on the device energy efficiency parameters in the three-dimensional manifold coordinate model, the specific process of constructing the energy scalar field using the Gaussian kernel density estimation method is as follows: first, determine the energy efficiency state of each coordinate point in the three-dimensional manifold space, and assign a positive energy efficiency value to the high-efficiency region coordinate point and a negative energy efficiency value to the low-efficiency region coordinate point. The Gaussian kernel density estimation method establishes a Gaussian kernel function distribution in the three-dimensional manifold space with each coordinate point as the center, and the bandwidth parameter of the kernel function is automatically optimized and determined by the Silverman criterion. The energy density distribution of the entire three-dimensional manifold space is calculated by superimposing the Gaussian kernel function contributions of all coordinate points. The energy scalar field valuation process considers the spatial correlation of adjacent coordinate points, and the Gaussian kernel contributions of the high-efficiency region enhance each other to form a positive value region, and the Gaussian kernel contributions of the low-efficiency region superimpose each other to form a negative value region. The finally generated continuous energy scalar field completely characterizes the energy efficiency characteristics of each region in the three-dimensional manifold space, and the positive value region represents the high-efficiency region and the negative value region represents the low-efficiency region.

[0071] It should be noted that the main significance of assigning positive values to high-efficiency areas and negative values to low-efficiency areas is to establish an intuitive energy efficiency evaluation system. Positive values directly represent areas that meet or exceed energy efficiency standards, while negative values represent low-efficiency areas that do not meet standards. This polar assignment makes the energy efficiency state visualized, with zero values naturally forming a dividing line. At the same time, the numerical size reflects the difference in energy efficiency level, facilitating quantitative analysis and optimization direction judgment. For example: +0.7 represents excellent energy efficiency, and -0.3 represents slight inefficiency.

[0072] The device energy efficiency parameters refer to real-time monitoring of voltage harmonic distortion rate, pipeline temperature gradient, storage tank pressure fluctuation, power factor, entropy increase rate, and frequency fluctuation in industrial plant areas.

[0073] Based on the energy scalar field, the power factor field and the entropy increase rate change are mapped into the equivalent energy quality field through convolution integration, and the expression is:

[0074]

[0075] where Φ(x, y, z) represents the equivalent energy quality field at (x, y, z) in the three-dimensional manifold coordinate model, θ is the time derivative of the entropy increase rate, α is the weight coefficient of the power factor field, β is the weight coefficient of the time derivative of the entropy increase rate, is the Green function kernel, * represents the convolution operator, and φ is the power factor field.

[0076] Further, based on the energy scalar field, the power factor field and the entropy increase rate change are combined by weighting to form a new field distribution. The Green function kernel is convolved with this combined field in three-dimensional space, and the convolution process integrates the surrounding area at each location in the manifold coordinate. Through this spatial convolution operation, the local characteristics of the power factor field and the entropy increase rate change are converted into an equivalent energy quality field with global correlation. Convolution integration combines the immediate energy efficiency state represented by the power factor with the energy efficiency trend reflected by the entropy increase rate change, generating a continuously distributed equivalent energy quality field in three-dimensional manifold space. The final equivalent energy quality field contains both spatial position information and energy quality characteristics, and can reflect the spatial correlation of different areas in energy efficiency performance.

[0077] Through the energy field reconstruction and geometric optimization method, a curved manifold space is generated and the optimal transmission path and twist intensity cloud map are identified, with the following steps,

[0078] Using the equivalent energy quality field Φ(x, y, z), construct the Riemannian metric tensor and establish the geometric structure of the energy space;

[0079] Further, the gradient vector of the equivalent energy mass field is calculated at each position point of the three-dimensional manifold coordinate model. The gradient vector represents the change rate of the equivalent energy mass field in each direction in space, reflecting the local variation characteristics of the energy distribution. A symmetric positive definite second-order covariant tensor is constructed using the gradient vector as the basis form of the Riemann metric tensor. The components of the Riemann metric tensor are determined by the first-order partial derivatives of the equivalent energy mass field, ensuring the positive definiteness of the tensor matrix. The distance metric between any two points in the energy space is defined by the Riemann metric tensor, establishing an energy space structure with clear geometric meaning. The geometric properties of the energy space, including curvature, connection, and other differential geometric properties, are completely determined by the Riemann metric tensor. The final energy space geometry maintains topological consistency with the original equivalent energy mass field.

[0080] The connection coefficients are obtained through the Christoffel symbol, and the Riemann curvature tensor is generated to form the curved manifold space;

[0081] Further, based on the established Riemann metric tensor, the Christoffel symbol is first calculated by the partial derivative of the metric tensor. The Christoffel symbol represents the parallel movement of the connection characteristics in the manifold space. The Riemann curvature tensor is constructed using the first-order derivative of the Christoffel symbol, which describes the curvature of the manifold space in any direction. Finally, the curved manifold space is generated while strictly maintaining the geometric correspondence with the original equivalent energy mass field, and its curvature characteristics directly reflect the spatial variation law of the energy distribution.

[0082] Based on the historical energy field evolution data, the initial flow direction of the equivalent energy mass field in the curved manifold space is defined, and the Eikonal equation is solved by the fast marching method to identify the optimal transmission path of the equivalent energy mass field in the curved manifold space, expressed as:

[0083]

[0084] where γ(t) is the optimal transmission path of the parameterized equivalent energy mass field in the curved manifold space, t is the parameter variable representing the path curve γ, γ represents the parameterized curve of the transmission path, g ij is the Riemann curvature tensor, d i is the differential component of the path curve γ in the i-th direction in the curved manifold space, d j is the differential component of the path curve γ in the j-th direction in the curved manifold space.

[0085] Further, based on the historical energy field evolution data, the initial flow direction of the equivalent energy mass field in the curved manifold space is first determined. The fast marching method establishes a wave front propagation model in the curved manifold space, and the minimum arrival time of each spatial position is determined through iteration. In the propagation process, the geometric characteristics of the curved manifold space directly affect the wave front expansion pattern, so that the distance metric naturally adapts to the spatial curvature characteristics. The identification of the optimal transmission path is realized by finding the curve that minimizes the path integral, which has the shortest path characteristic in the curved manifold space. The final optimal transmission path completely characterizes the best transmission trajectory of the equivalent energy mass field from the starting point to the terminal point.

[0086] By finite difference method, the intensity variation characteristics of the equivalent energy mass field at different positions are extracted, and based on Sobel operator, the energy gradient is generated;

[0087] Further, based on the historical energy field evolution data, the initial flow direction of the equivalent energy mass field in the curved manifold space is first determined. The fast marching method establishes a wave front propagation model in the curved manifold space, and the minimum arrival time of each spatial position is determined through iteration. In the propagation process, the geometric characteristics of the curved manifold space directly affect the wave front expansion pattern, so that the distance metric naturally adapts to the spatial curvature characteristics. The identification of the optimal transmission path is realized by finding the curve that minimizes the path integral, which has the shortest path characteristic in the curved manifold space. The final optimal transmission path completely characterizes the best transmission trajectory of the equivalent energy mass field from the starting point to the terminal point.

[0088] The Riemannian geometric curvature and gradient coupling algorithm is used to calculate the twist intensity of the optimal transmission path, and the expression is:

[0089]

[0090] wherein, represents the local twist intensity of each point along the optimal transmission path γ(t), and R(γ(t)) is the Riemannian curvature tensor on the optimal transmission path, represents the gradient vector of the energy scalar field at each point on the optimal transmission path.

[0091] Further, on the optimal transmission path, the path twist intensity distribution is generated by coupling the Riemannian geometric curvature and the energy gradient characteristics. First, the spatial curvature characteristics of each point in the optimal transmission path are extracted, which quantifies the geometric bending degree of the manifold; at the same time, the gradient variation characteristics of the energy scalar field along the path are obtained, which represent the energy distribution change rate. The curvature characteristics and gradient characteristics are coupled and analyzed, the curvature reflects the spatial deformation, and the gradient reflects the energy fluctuation, both of which determine the path twist state. The finally generated optimal transmission path twist intensity distribution completely quantifies the geometric and physical comprehensive twist degree of each position of the path. The maximum value of the twist intensity corresponds to the section in the path that is most significantly affected by the spatial bending and energy change, and the minimum value corresponds to the most stable transmission section.

[0092] It should be noted that the curvature feature refers to the average curvature at each point in the curved manifold space, which is used to quantify the local bending degree of the manifold surface; the gradient feature refers to the rate of change of the equivalent energy quality field at each point in the manifold space, which is used to represent the local change trend and intensity of the energy distribution.

[0093] Based on the statistical distribution of historical data, the distortion intensity threshold w is defined;

[0094] When , it is considered that the current region belongs to the energy efficiency black hole region;

[0095] When , it is considered that the current region belongs to the energy efficiency black surface, which is the critical interface between the energy efficiency black hole region and the normal energy efficiency region;

[0096] When , it is considered that the current region has a normal energy efficiency state;

[0097] It should be noted that based on the historical energy field evolution data, the specific process of defining the distortion intensity threshold w is as follows: first, collect all the local distortion intensity data samples on the historical optimal transmission path. The kernel density estimation method is used to construct the probability density distribution curve of the distortion intensity. According to the probability density distribution curve of the distortion intensity, the distortion intensity values corresponding to different quantiles are investigated, and the 85-95 percentile interval is focused on. Combined with the inflection point analysis of energy transmission efficiency, the critical value that can cover typical working conditions and has engineering significance is determined. The final value of the distortion intensity threshold w is determined through repeated sampling verification. For example: when the analysis shows that 95% of the sample distortion intensity is less than 0.5, and the energy transmission efficiency decreases significantly near 0.5, the distortion intensity threshold w is defined as 0.5.

[0098] Based on the identified energy efficiency black hole region, a distortion intensity cloud map is generated by the Marching Cubes algorithm.

[0099] Further, based on the identified energy efficiency black hole region, the Marching Cubes algorithm processes three-dimensional distortion intensity data to generate a visual cloud map. First, a three-dimensional grid structure containing the energy efficiency black hole region is established, and each grid vertex is assigned a corresponding distortion intensity value. The algorithm scans each voxel unit, calculates the intersection points of the isosurface and the unit edge by linear interpolation. According to the intersection point distribution mode, match the preset triangulation scheme, and construct triangular facets in the unit that approximate the isosurface. After all units are processed, connect the triangular facets of adjacent units to form a continuous curved surface. The final output distortion intensity cloud map accurately presents the spatial distribution characteristics of the energy efficiency black hole region. The color of the curved surface in the cloud map maps the distortion intensity value, and the gradient change reflects the degree of energy transmission obstruction, realizing the intuitive and visual expression of distortion intensity in three-dimensional space.

[0100] S4, generating a thermal color scale based on the distortion intensity cloud map through a thermal color scale mapping algorithm, and rendering a three-dimensional gradient field distribution of the equivalent energy quality field using a ray tracing method to generate a visual energy evaluation interface.

[0101] Based on the distortion intensity cloud map, a thermal color scale is generated from cold to warm through a thermal color scale mapping algorithm;

[0102] Further, based on the distortion intensity cloud map, the thermal color scale mapping algorithm converts the distortion intensity values into a continuous color scale from cold to warm. First, determine the minimum and maximum values of the distortion intensity value range to establish a complete numerical interval. Define the starting cold hue and the ending warm hue in the HSV color space, usually choose blue as the cold end point and red as the warm end point. The thermal color scale mapping algorithm constructs a smooth transition color scale curve in the color space, ensuring the visual continuity of adjacent color scales. Each distortion intensity value corresponds to a specific color on the color scale curve through linear mapping, the smaller the value, the colder the hue, the larger the value, the warmer the hue. The color mapping process considers the perception characteristics of the human eye to the color change, appropriately increases the color scale density in the transition area. The final generated continuous color scale accurately reflects the numerical distribution characteristics of the distortion intensity cloud map, the cold color area represents low distortion intensity, the warm color area represents high distortion intensity, and the intermediate transition color scale corresponds to medium intensity values.

[0103] Based on the continuous color scale, the equivalent energy quality field is colored using a color scale-based optical property conversion method;

[0104] Further, based on the continuous color scale, the specific process of coloring the equivalent energy quality field using the color scale-based optical property conversion method is as follows: first, establish a strict correspondence between the numerical range of the equivalent energy quality field and the continuous color scale to ensure that each numerical interval is mapped to a specific color scale interval. In the RGB color space, smooth conversion of numerical values to colors is achieved through a bilinear interpolation algorithm, maintaining the continuity of color transition. The optical property conversion process considers the visual perception characteristics of the human eye to adjust the color saturation and brightness in the key numerical interval, enhancing the visual contrast effect. The numerical value of each spatial point in the equivalent energy quality field obtains the corresponding RGB color value by looking up the color scale mapping table, the low energy area is mapped to the cold hue, and the high energy area is mapped to the warm hue. The colored equivalent energy quality field forms a three-dimensional color distribution with optical depth, and the color gradient accurately reflects the spatial variation characteristics of the energy field.

[0105] The colored equivalent energy quality field is used as a non-uniform participating medium to generate a pseudo-color volume rendering image through a Monte Carlo path tracing algorithm;

[0106] Further, first, a three-dimensional space grid structure containing the colored equivalent energy quality field is established, and each grid cell stores the corresponding RGB color value and optical characteristic parameter. The Monte Carlo path tracking algorithm emits random sampling light rays under the virtual camera perspective, and the light rays interact with the grid cells when propagating in the non-uniform medium. During the light propagation process, the absorption and scattering effects of the light energy are calculated according to the color value and optical parameters of the grid cell, and the Russian roulette method is used to determine the termination or continuous propagation of the light. The statistical average of multiple sampling results forms the final color value of each pixel, and the cool tone of the low-energy area and the warm tone of the high-energy area produce a natural transition in the rendering process. The final generated pseudo-color volume rendering image accurately presents the three-dimensional spatial distribution characteristics of the equivalent energy quality field, and the color change reflects the gradient change of the energy intensity. The internal structure is expressed in levels through transparency adjustment. The rendering result maintains geometric consistency with the original data, while enhancing the visual recognition of the energy distribution characteristics.

[0107] The pseudo-color volume rendering image, the energy efficiency black hole isosurface, and the three-dimensional grid space are fused through FBO to generate a visual energy evaluation interface.

[0108] Further, the multi-layer fusion process based on FBO first renders the pseudo-color volume rendering image, the energy efficiency black hole isosurface, and the three-dimensional grid space to different color attachments respectively. The pseudo-color volume rendering image provides the overall distribution characteristics of the equivalent energy quality field, the energy efficiency black hole isosurface highlights the key abnormal areas, and the three-dimensional grid space maintains the basic geometric structure. Multi-layer mixing is realized in the fragment shader, and the alpha blending algorithm is used to control the transparency and superimposition order of each layer. The energy efficiency black hole isosurface improves the visual recognition through edge enhancement processing, and the three-dimensional grid space is rendered in wireframe mode to maintain structure visibility. Color space conversion is considered in the mixing process to ensure the consistency of color expression of different layers. The final generated visual energy evaluation interface presents the energy distribution of the pseudo-color volume rendering image, the abnormal area of the energy efficiency black hole isosurface, and the geometric reference of the three-dimensional grid space. Each layer of information realizes coordinated and unified visualization expression through reasonable transparency matching.

[0109] The embodiment also provides a smart energy multi-energy evaluation visualization system for an industry, comprising a data acquisition module, a three-dimensional manifold generation module, a cloud map generation module and a visualization module; the data acquisition module is used for acquiring multi-source heterogeneous data and performing preprocessing, generating high-dimensional energy data aligned in time and space through a time-space hash coding method; the three-dimensional manifold generation module is used for performing local neighborhood optimization and global topology preservation through a UMAP manifold learning algorithm, and mapping the high-dimensional energy data aligned in time and space into a three-dimensional manifold coordinate model with energy efficiency labels; the cloud map generation module is used for converting device energy efficiency parameters into an equivalent energy quality field based on the three-dimensional manifold coordinate model, generating a curved manifold space and identifying an optimal transmission path and a distortion intensity cloud map through an energy field reconstruction and geometric optimization method; and the visualization module is used for generating a thermal color scale through a thermal color scale mapping algorithm based on the distortion intensity cloud map, rendering a three-dimensional gradient field distribution of the equivalent energy quality field by using a ray tracing method, and generating a visual energy evaluation interface.

[0110] The embodiment also provides a computer device suitable for the smart energy multi-energy evaluation visualization method for an industry, comprising a memory and a processor; the memory is used for storing computer executable instructions, and the processor is used for executing the computer executable instructions to realize the smart energy multi-energy evaluation visualization method for an industry as proposed in the above embodiment.

[0111] The computer device can be a terminal, and the computer device comprises a processor, a memory, a communication interface, a display screen and an input device connected through a system bus. The processor of the computer device is used for providing computing and control capabilities. The memory of the computer device comprises a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operating system and the computer program in the non-volatile storage medium. The communication interface of the computer device is used for wired or wireless communication with external terminals. The wireless communication can be achieved through WIFI, an operator network, NFC (near field communication) or other technologies. The display screen of the computer device can be a liquid crystal display screen or an electronic ink display screen. The input device of the computer device can be a touch layer overlaid on the display screen, or can be a key, trackball or touchpad arranged on the shell of the computer device, or can be an external keyboard, touchpad or mouse, etc.

[0112] The embodiment also provides a storage medium on which a computer program is stored, the program being executed by a processor to implement the method for visualizing multi-energy evaluation of smart energy for industry as proposed in the above embodiment; the storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic memory, flash memory, magnetic disk or optical disk.

[0113] To sum up, the application achieves accurate spatio-temporal alignment of multi-source heterogeneous data by generating space-time hash codes through a spatio-temporal hash coding method and establishing a device-grid mapping relationship in combination with GeoHash coding, and provides a high-precision data basis for subsequent analysis; the application reveals the essential distribution characteristics of energy data by mapping high-dimensional data into three-dimensional manifold coordinates that maintain the global topological structure through Minkowski distance measurement and random gradient descent optimization. The two steps cooperatively build a complete analysis chain from data preprocessing to spatial mapping, making the energy efficiency evaluation result more accurate and reliable, and providing key technical support for intelligent management of industrial energy systems.

[0114] It should be noted that the above embodiments are only used to illustrate the technical solutions of the application rather than limit the application. Although the application has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the application can be modified or replaced equivalently without departing from the spirit and scope of the technical solutions of the application, and all of them should be covered in the scope of the claims of the application.

Claims

1. A visualization method for intelligent energy multi-energy evaluation in industry, characterized in that: include, Multi-source heterogeneous data is collected and preprocessed. Spatiotemporal hashing encoding method is used to generate spatiotemporally aligned high-dimensional energy data. The multi-source heterogeneous data includes voltage harmonic distortion rate, pipeline temperature gradient, tank pressure fluctuation, power factor, entropy increase rate and frequency fluctuation. The UMAP manifold learning algorithm is used to perform local neighborhood optimization and global topology preservation, and to map spatiotemporally aligned high-dimensional energy data into a three-dimensional manifold coordinate model with energy efficiency labels. Based on the three-dimensional manifold coordinate model, the energy efficiency parameters of the equipment are transformed into an equivalent energy mass field. Through energy field reconstruction and geometric optimization methods, a curved manifold space is generated and the optimal transmission path and torsion intensity cloud map are identified. Based on the distortion intensity cloud map, a thermal color scale is generated through a thermal color scale mapping algorithm, and a ray tracing method is used to render the three-dimensional gradient field distribution of the equivalent energy mass field, generating a visualized energy evaluation interface.

2. The intelligent energy multi-energy evaluation visualization method for industry as described in claim 1, characterized in that: The preprocessing includes denoising, outlier removal, and sliding window smoothing of multi-source heterogeneous data. The steps for generating spatiotemporally aligned high-dimensional energy data using a spatiotemporal hashing encoding method are as follows: The physical space of the industrial plant area is divided into a three-dimensional grid space using a spatial discretization algorithm; Spatial hash codes are generated using the MD5 hash algorithm, and timestamp hash codes are generated using a time window and CRC32 algorithm. The GeoHash encoding algorithm is used to establish a device-grid mapping relationship. Based on the device-grid mapping relationship, the spatial hash code and timestamp hash code are spatiotemporally aligned to generate spatiotemporally aligned high-dimensional energy data.

3. The intelligent energy multi-energy evaluation visualization method for industry as described in claim 2, characterized in that: The steps for local neighborhood optimization and global topology preservation using the UMAP manifold learning algorithm are as follows: The Euclidean distance matrix of high-dimensional energy data is obtained by using the Minkowski distance metric method, and the Euclidean distance matrix is ​​converted into a probability distribution map using an exponential kernel function. An initial set of coordinate points is randomly generated in a three-dimensional grid space using a uniformly distributed random sampling method. Using the initial set of coordinate points as the starting point of the optimization process, the coordinates of the three-dimensional mesh cells are iteratively optimized by minimizing the cross-entropy loss between the high-dimensional probability graph and the low-dimensional embedding through the stochastic gradient descent algorithm, thereby generating three-dimensional manifold coordinates with global topology preservation properties.

4. The intelligent energy multi-energy evaluation visualization method for industry as described in claim 3, characterized in that: The process of mapping spatiotemporally aligned high-dimensional energy data into a three-dimensional manifold coordinate model with energy efficiency labels refers to defining an entropy increase rate threshold, and based on the three-dimensional manifold coordinates, combined with the power factor and the entropy increase rate threshold, using SVM to divide the energy efficiency region into high-efficiency and low-efficiency regions, and assigning energy efficiency labels to them, thereby generating a manifold coordinate model containing three-dimensional coordinates and energy efficiency labels.

5. The intelligent energy multi-energy evaluation visualization method for industry as described in claim 1, characterized in that: The process of converting equipment energy efficiency parameters into an equivalent energy mass field refers to mapping the equipment energy efficiency parameters in the three-dimensional manifold coordinate model into an energy scalar field using the Gaussian kernel density estimation method, assigning values ​​to the energy efficiency region, and then mapping the power factor field and entropy increase rate change into an equivalent energy mass field through convolution integral based on the energy scalar field.

6. The intelligent energy multi-energy evaluation visualization method for industry as described in claim 1, characterized in that: The process of generating a curved manifold space and identifying the optimal transmission path and torsion intensity contour map through energy field reconstruction and geometric optimization methods is as follows: Using the equivalent energy-mass field, a Riemannian metric tensor is constructed and the geometric structure of the energy space is established; The connection coefficients are obtained through Christoffel notation and combined with the Riemann curvature tensor to generate the curved manifold space; The initial flow direction of the equivalent energy mass field in the curved manifold space is defined, and the optimal transport path of the equivalent energy mass field in the curved manifold space is identified by solving the Eikonal equation using the fast travel method. The Riemannian geometric curvature and gradient coupling algorithm is used to calculate the torsion intensity of the optimal transmission path; Define a torsion intensity threshold and compare it with the torsion intensity of the optimal transmission path to identify energy-efficient black hole regions; Based on the identified energy efficiency black hole regions, a distortion intensity cloud map is generated using the Marching Cubes algorithm.

7. The intelligent energy multi-energy evaluation visualization method for industry as described in claim 6, characterized in that: The process involves generating thermal color levels using a thermal color level mapping algorithm and rendering the three-dimensional gradient field distribution of the equivalent energy mass field using ray tracing to create a visualized energy evaluation interface. The steps are as follows: Based on the distortion intensity cloud map, a continuous color scale from cool to warm colors is generated by the thermal color scale mapping algorithm. An optical property conversion method based on color levels is used to color the equivalent energy mass field; The colored equivalent energy mass field is used as a non-uniform participating medium, and a pseudo-color volume rendering image is generated by the Monte Carlo path tracing algorithm. By using FBO, pseudo-color volume rendering images, energy efficiency black hole isosurfaces, and three-dimensional mesh space are fused into multiple layers to generate a visualized energy evaluation interface.

8. A smart energy multi-energy evaluation visualization system for industry, based on the smart energy multi-energy evaluation visualization method for industry according to any one of claims 1 to 7, characterized in that: It includes a data acquisition module, a 3D manifold generation module, a cloud map generation module, and a visualization module; The data acquisition module is used to collect multi-source heterogeneous data and preprocess it, and generate spatiotemporally aligned high-dimensional energy data through the spatiotemporal hashing encoding method. The 3D manifold generation module is used to perform local neighborhood optimization and global topology preservation through the UMAP manifold learning algorithm, and to map spatiotemporally aligned high-dimensional energy data into a 3D manifold coordinate model with energy efficiency labels. The cloud map generation module is used to convert equipment energy efficiency parameters into an equivalent energy mass field based on a three-dimensional manifold coordinate model. Through energy field reconstruction and geometric optimization methods, it generates curved manifold space and identifies the optimal transmission path and torsion intensity cloud map. The visualization module is used to generate thermal color levels based on the distortion intensity cloud map through a thermal color level mapping algorithm, and to render the three-dimensional gradient field distribution of the equivalent energy mass field using ray tracing, thereby generating a visualized energy evaluation interface.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that: When the processor executes the computer program, it implements the steps of the intelligent energy multi-energy evaluation visualization method for industry as described in any one of claims 1 to 7.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by the processor, it implements the steps of the intelligent energy multi-energy evaluation visualization method for industry as described in any one of claims 1 to 7.

Images