Electric actuator energy consumption optimization method and system based on big data analysis

Abstract

The invention discloses an electric actuator energy consumption optimization method and system based on big data analysis, and particularly relates to the technical field of energy consumption optimization. Multi-source operation data are collected, a sliding window is utilized to extract time sequence features to generate a working condition modal feature vector, and an energy efficiency influence factor set is constructed in combination with an energy efficiency influence factor weight matrix. Performing working condition state clustering on the energy efficiency influence factor set by using a dynamic density clustering algorithm to generate an operation mode classification tree; and performing energy consumption traceability analysis on the unsteady operation mode to generate an energy efficiency degradation characteristic spectrum. Inputting the energy efficiency degradation characteristic spectrum into a pre-trained energy consumption optimization decision model, outputting a dynamic regulation strategy set, and outputting a final optimization control parameter packet by comparing the deviation degree of an energy efficiency transition track and a theoretical optimal curve and iteratively updating the weight of the model after the strategy takes effect. And intelligent optimization and energy-saving control of the operation mode of the electric actuator are realized.

Description

Technical Field

[0001] The present invention relates to the technical field of energy consumption optimization, and more specifically, to a method and system for optimizing energy consumption of an electric actuator based on big data analysis. Background Art

[0002] Traditional electric actuator energy management methods usually rely on static parameter settings and simple control strategies, which are difficult to cope with energy consumption fluctuations under variable working conditions and lack full utilization of real-time operating data. Existing technologies have deficiencies in data fusion, time series feature extraction, and working condition analysis, and cannot accurately characterize the complex relationship between the actuator operating mode and energy efficiency attenuation, resulting in delayed response of the control strategy and limited energy consumption optimization effect.

[0003] In order to solve the above problems, a technical solution is now provided. Summary of the invention

[0004] In order to overcome the above-mentioned defects of the prior art, an embodiment of the present invention provides an electric actuator energy consumption optimization method and system based on big data analysis to solve the problems raised in the above-mentioned background technology.

[0005] To achieve the above object, the present invention provides the following technical solutions: The energy consumption optimization method of electric actuator based on big data analysis includes the following steps: Collect multi-source operation data streams of electric actuators, extract time series features of the multi-source operation data streams based on sliding windows, and generate actuator operating mode feature vectors; According to the actuator working condition modal eigenvector and the preset energy efficiency influencing factor weight matrix, an energy efficiency influencing factor set is constructed; Based on the dynamic density clustering algorithm, the energy efficiency influencing factor set is clustered into working conditions to generate the actuator operation mode classification tree; Conduct energy consumption source analysis on non-steady-state operation modes and generate energy efficiency degradation characteristic maps; Input the energy efficiency degradation feature map into the pre-trained energy consumption optimization decision model and output a dynamic control strategy set; After the dynamic control strategy set takes effect, the weight distribution of the energy consumption optimization decision model is iteratively updated by comparing the deviation between the energy efficiency transition trajectory before and after the implementation of the strategy and the theoretical optimal energy consumption curve, and the final optimization control parameter package is output.

[0006] In a preferred embodiment, the multi-source operation data stream of the electric actuator is collected, and the time series feature extraction of the multi-source operation data stream is performed based on the sliding window to generate the actuator working condition modal feature vector, specifically: Collect the motor current signal, valve position feedback signal, control instruction sequence and vibration spectrum data of the electric actuator to form a multi-source heterogeneous operation data stream; Adjust the size of the sliding window according to the frequency of changes in the control instruction sequence; Extract time series features from multi-source heterogeneous operation data streams within a sliding window; The time domain features, frequency domain features and wavelet transform features extracted in each sliding window are dimensionally spliced ​​to generate the actuator working condition modal feature vector.

[0007] In a preferred embodiment, according to the actuator working condition modal characteristic vector, combined with the preset energy efficiency influencing factor weight matrix, an energy efficiency influencing factor set is constructed, specifically: Extracting time domain characteristic parameters, frequency domain characteristic parameters and wavelet transform characteristic parameters from the actuator working mode characteristic vector; Match the row vectors of the energy efficiency impact factor weight matrix with the characteristic parameters of the corresponding type respectively; Performing a weighted calculation operation on each feature parameter; The weighted calculation results are accumulated according to the feature type to generate a multi-dimensional energy efficiency influencing factor set.

[0008] In a preferred embodiment, the energy efficiency influencing factor set is clustered according to the operating state based on the dynamic density clustering algorithm to generate an actuator operation mode classification tree, specifically: The factor whose product of local density and relative distance in the multi-dimensional energy efficiency influencing factor concentration exceeds the threshold is selected as the initial cluster center; According to the initial cluster center distribution density, the continuous operating conditions with similar density and continuous distribution of energy efficiency influencing factors are divided into the same cluster branch; The discrete factors that meet the density connectivity conditions are classified into adjacent cluster branches; A tree classification structure is constructed, and pruning optimization is performed based on the mean and variance of the energy efficiency index of the clustering branches to generate an actuator operation mode classification tree including steady-state operation modes and non-steady-state operation modes.

[0009] In a preferred embodiment, energy consumption source analysis is performed on the non-steady-state operation mode to generate an energy efficiency degradation characteristic map, specifically: Identify the operating state range of the non-steady-state operation mode and determine the threshold range of the energy efficiency influencing factor; According to the threshold range of energy efficiency influencing factors, filter out instruction segments with abnormally increased energy consumption, extract abnormal pulse patterns related to the control instruction sequence, and mark the start and end times of the abnormal pulse patterns; Taking the start and end time of the abnormal pulse mode as a reference, calculate the abnormal change amount of the electric actuator motor current signal and the hysteresis response amount of the valve position feedback signal; Based on the numerical correlation between the abnormal change of the motor current signal and the hysteresis response of the valve position feedback signal, an energy efficiency degradation characteristic map is generated.

[0010] In a preferred embodiment, the energy efficiency degradation feature map is input into a pre-trained energy consumption optimization decision model, and a dynamic control strategy set is output, specifically: A training set of energy efficiency degradation feature maps of electric actuators is constructed based on historical operation data, and an energy consumption optimization decision model is trained through supervised learning. Normalizing the characteristic parameters of the energy efficiency degradation characteristic map to obtain a standardized characteristic input matrix; The standardized feature input matrix is ​​input into the pre-trained energy consumption optimization decision model to perform feature matching and energy efficiency pattern recognition; According to the feature matching and energy efficiency pattern recognition results, a dynamic control strategy set corresponding to the operating mode of the electric actuator is output.

[0011] In a preferred embodiment, after the dynamic control strategy set takes effect, the weight distribution of the energy consumption optimization decision model is iteratively updated by comparing the deviation between the energy efficiency transition trajectory before and after the implementation of the strategy and the theoretical optimal energy consumption curve, and the final optimization control parameter package is output, specifically: Real-time record of the change trend of the motor current signal, valve position feedback signal and vibration spectrum data of the electric actuator, and draw the actual energy efficiency transition trajectory; Compare the actual energy efficiency transition trajectory with the theoretical optimal energy consumption curve point by point, and calculate the degree of deviation between the actual energy efficiency transition trajectory and the theoretical optimal energy consumption curve; The deviation degree is fed back to the energy consumption optimization decision model as an error, and the weight distribution within the energy consumption optimization decision is iteratively adjusted based on the error back propagation mechanism; When the degree of deviation drops to within the preset allowable range, the final optimized control parameter package is output.

[0012] On the other hand, the present invention provides an electric actuator energy consumption optimization system based on big data analysis, including a feature extraction module, an influencing factor construction module, a working condition clustering module, an energy consumption tracing analysis module, a control strategy output module, and a model updating module; The feature extraction module collects the multi-source operation data stream of the electric actuator, extracts the time series features of the multi-source operation data stream based on the sliding window, and generates the actuator working condition modal feature vector; The influence factor construction module constructs an energy efficiency influence factor set based on the actuator working condition modal eigenvector and the preset energy efficiency influence factor weight matrix; The working condition clustering module clusters the working condition of the energy efficiency influencing factor set based on the dynamic density clustering algorithm and generates an actuator operation mode classification tree; The energy consumption source analysis module performs energy consumption source analysis on non-steady-state operation modes and generates energy efficiency degradation characteristic maps; The control strategy output module inputs the energy efficiency degradation feature map into the pre-trained energy consumption optimization decision model and outputs a dynamic control strategy set; After the dynamic control strategy set takes effect, the model update module iteratively updates the weight distribution of the energy consumption optimization decision model by comparing the deviation between the energy efficiency transition trajectory before and after the implementation of the strategy and the theoretical optimal energy consumption curve, and outputs the final optimization control parameter package.

[0013] Technical effects and advantages of the electric actuator energy consumption optimization method and system based on big data analysis of the present invention: 1. Real-time acquisition of multi-source operating data, and use of sliding window technology to extract time series features to accurately reflect the working conditions of electric actuators; combined with the preset energy efficiency influencing factor weight matrix, the extracted time domain, frequency domain and wavelet transform features are weighted and accumulated to construct a multi-dimensional energy efficiency influencing factor set; the energy efficiency influencing factor set is divided into working conditions using a dynamic density clustering algorithm to form a classification tree containing steady-state and non-steady-state operating modes; in the non-steady-state mode, abnormal pulse modes are identified through energy consumption source tracing analysis, and the coupling relationship between energy efficiency attenuation and response lag is quantified to generate an energy efficiency degradation feature map; 2. Input the energy efficiency degradation feature map into the pre-trained decision model to output the dynamic control strategy. After the strategy takes effect, compare the deviation between the actual energy efficiency transition trajectory and the theoretical optimal curve, use error feedback to iteratively update the model weight, and output the final optimized control parameter package. Effectively improve the operating efficiency and energy consumption management accuracy of electric actuators, reduce equipment energy consumption and operating costs, enhance adaptive capabilities, and provide reliable technical support for intelligent manufacturing and green energy conservation. BRIEF DESCRIPTION OF THE DRAWINGS

[0014] Figure 1 It is a schematic diagram of the energy consumption optimization method of the electric actuator based on big data analysis of the present invention; Figure 2 It is a structural schematic diagram of the electric actuator energy consumption optimization system based on big data analysis of the present invention. DETAILED DESCRIPTION

[0015] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0016] Embodiment 1:

[0017] Figure 1 The present invention provides an electric actuator energy consumption optimization method based on big data analysis, which includes the following steps: Collect multi-source operation data streams of electric actuators, extract time series features of the multi-source operation data streams based on sliding windows, and generate actuator operating mode feature vectors; According to the actuator working condition modal eigenvector and the preset energy efficiency influencing factor weight matrix, an energy efficiency influencing factor set is constructed; Based on the dynamic density clustering algorithm, the energy efficiency influencing factor set is clustered into working conditions to generate the actuator operation mode classification tree; Conduct energy consumption source analysis on non-steady-state operation modes and generate energy efficiency degradation characteristic maps; Input the energy efficiency degradation feature map into the pre-trained energy consumption optimization decision model and output a dynamic control strategy set; After the dynamic control strategy set takes effect, the weight distribution of the energy consumption optimization decision model is iteratively updated by comparing the deviation between the energy efficiency transition trajectory before and after the implementation of the strategy and the theoretical optimal energy consumption curve, and the final optimization control parameter package is output.

[0018] Specifically, the multi-source operation data stream of the electric actuator is collected, and the time series features of the multi-source operation data stream are extracted based on the sliding window to generate the actuator working condition modal feature vector, including: Collect the motor current signal, valve position feedback signal, control instruction sequence and vibration spectrum data of the electric actuator to form a multi-source heterogeneous operation data stream; Among them, the motor current signal indicates the current consumed by the electric actuator driving motor during operation, reflecting the load conditions; the valve position feedback signal indicates the actual feedback information of the valve position control in the electric actuator, reflecting the actual degree of opening or closing; the control instruction sequence indicates a series of instructions sent to the electric actuator, such as start, stop, speed regulation, etc., which usually come with timestamps; the vibration spectrum data refers to the spectrum data obtained by performing fast Fourier transform on the vibration signal collected by the vibration sensor, which can be used to monitor abnormal vibration or fault signs.

[0019] For example, assume that in a certain acquisition process: the motor current signal is sampled 1000 times per second to obtain 1000 data points within 1 second; the valve position feedback signal is sampled 100 times per second; the control instruction sequence may have only 3 to 5 instructions within a second; the vibration spectrum data is calculated by fast Fourier transform, and each calculation generates a vector containing 50 frequency band amplitudes. After aligning these data according to their respective sampling times, a multi-source heterogeneous operation data stream is formed; Adjust the size of the sliding window according to the frequency of changes in the control instruction sequence; Specifically, the sliding window is used to divide the data in a continuous time period for local feature extraction. When the control instruction frequency is high (for example, the instruction is updated every 50 milliseconds), a shorter sliding window (for example, 200 milliseconds) is required to capture the rapidly changing information.

[0020] When the control instruction frequency is low (for example, the instruction is updated every 500 milliseconds), a longer sliding window (for example, 1 second or longer) can be used to fully extract features.

[0021] For example, if the control instruction update frequency in a certain operation segment is once every 100 milliseconds, the sliding window can be set to 500 milliseconds, that is, each window contains about 5 control instructions; if the update frequency is once every 300 milliseconds, the window can be set to 1.5 seconds to ensure that each window contains enough data for feature extraction.

[0022] Extract time series features from multi-source heterogeneous operation data streams within a sliding window; Specifically, for each sliding window, the time series data of each signal in the window is analyzed to extract features that can describe the law of data change, including: Time domain characteristics: such as mean value, standard deviation, maximum value, minimum value, etc.; Frequency domain characteristics: such as main frequency components, spectrum energy distribution, etc.; Wavelet transform features: Multi-scale components are extracted through wavelet transform, such as the energy changes of signals at different scales.

[0023] Exemplarily, within a 500 millisecond window, the average value of the motor current signal can be calculated, and Fourier transform can be performed to calculate the main frequency components; wavelet transform can be used to extract detail coefficients and approximate coefficients from the vibration signal.

[0024] The time domain features, frequency domain features and wavelet transform features extracted in each sliding window are dimensionally spliced ​​to generate the actuator working condition modal feature vector; Specifically, all features extracted in each sliding window are combined in a certain order to form a high-dimensional feature vector, which can fully reflect the operating status of the actuator in the window.

[0025] For example, the features extracted within a sliding window include: the average value, standard deviation, maximum value, and minimum value of the motor current; the main frequency and spectrum energy of the vibration signal; and energy of different scales extracted by wavelet transform.

[0026] These features are concatenated in a pre-set order to obtain the operating condition modal feature vector.

[0027] For example, assume that the features extracted in a window are as follows: Motor current signal: average value = 5A, standard deviation = 0.5A; Valve position feedback signal: average value = 30°, standard deviation = 2°; Vibration spectrum data: main frequency = 50Hz, spectrum energy = 200; Wavelet transform: scale 1 energy = 100, scale 2 energy = 80.

[0028] Then the feature vector of this window can be expressed as: [5, 0.5, 30, 2, 50, 200, 100, 80].

[0029] Specifically, according to the actuator working condition modal eigenvector and the preset energy efficiency influencing factor weight matrix, an energy efficiency influencing factor set is constructed, including: Extracting time domain characteristic parameters, frequency domain characteristic parameters and wavelet transform characteristic parameters from the actuator working mode characteristic vector; Specifically, the generated working condition modal feature vector contains information from different time domains and frequency domains. Different types of feature parameters in the vector are separated and used as feature parameters of time domain, frequency domain and wavelet transform respectively: Time domain characteristic parameters: such as mean value, standard deviation, extreme value, etc.; Frequency domain characteristic parameters: such as main frequency and spectrum energy; Wavelet transform characteristic parameters: such as energy distribution and detail coefficient at each scale.

[0030] For example, in the vector [5, 0.5, 30, 2, 50, 200, 100, 80], the first 4 numbers (5, 0.5, 30, 2) can be considered as time domain feature parameters, the next 2 numbers (50, 200) are frequency domain feature parameters, and the last 2 numbers (100, 80) are wavelet transform feature parameters. The row vectors of the energy efficiency impact factor weight matrix are matched with the characteristic parameters of the corresponding type respectively.

[0031] Specifically, in the preset energy efficiency impact factor weight matrix, each row corresponds to a characteristic parameter, and a weight is assigned to the characteristic parameter to reflect the importance of the characteristic to the energy efficiency.

[0032] For example, if the first row in the matrix corresponds to the mean value of the motor current, its weight may be 0.3; the second row corresponds to the standard deviation, its weight may be 0.2, and so on.

[0033] The matching process is to match each feature parameter extracted from the feature vector with the weight of the corresponding row in the weight matrix.

[0034] Perform a weighted calculation operation on each feature parameter; specifically, perform a product operation on each extracted feature parameter and its corresponding weight to obtain a weighted value; Specifically, according to the feature type (time domain, frequency domain, wavelet transform), the weighted values ​​are summed to generate a value that reflects the overall impact of the characteristics of this category, which is called the energy efficiency index.

[0035] The weighted calculation results are accumulated according to the feature type to generate a multi-dimensional energy efficiency influencing factor set.

[0036] Specifically, the energy efficiency influencing factor set is clustered based on the dynamic density clustering algorithm to generate the actuator operation mode classification tree, including: The factor whose product of local density and relative distance in the multi-dimensional energy efficiency influencing factor concentration exceeds the threshold is selected as the initial cluster center; Specifically, in the dynamic density clustering algorithm, commonly used indicators include local density (i.e., the point density around a data point) and relative distance (i.e., the distance between the point and points with higher density).

[0037] For each factor point, the local density can be defined as the number of factor points within the radius.

[0038] For each factor point, the relative distance is expressed as its distance to the nearest point with higher density than it.

[0039] If the product of the local density and the relative distance exceeds the preset threshold, the factor point meets the requirements and is selected as the initial cluster center.

[0040] According to the initial cluster center distribution density, the continuous operating conditions with similar density and continuous distribution of energy efficiency influencing factors are divided into the same cluster branch; Specifically, for the selected initial cluster centers, factors with similar density and continuous distribution are classified into the same branch according to their respective density values, that is, these factors are considered to represent the same working condition. For example, if the local density of two cluster centers is between 8 and 12, and the distance between them is small, they can be classified into the same cluster branch.

[0041] Assuming that in the energy efficiency influencing factor set, there are two initial cluster centers A and B, whose local densities are 9 and 10 respectively, and the relative distances are within the allowable range, then A and B and their nearby factors are classified as the same cluster branch.

[0042] The discrete factors that meet the density connectivity conditions are classified into adjacent cluster branches; Specifically, for those discrete factors that are not directly covered by the initial cluster center, if they meet the density connectivity with the factors in a cluster branch (that is, they can be connected through a series of adjacent high-density points), these discrete factors are classified into the cluster branch. For example, if the distance between a discrete factor and a point in a cluster branch is close and the local density of the discrete factor is high, it can be classified into the adjacent cluster branch through the connectivity rule.

[0043] Construct a tree classification structure, perform pruning optimization based on the mean and variance of the energy efficiency index of the clustering branches, and generate an actuator operation mode classification tree that includes steady-state operation modes and non-steady-state operation modes; Specifically, a tree structure is constructed based on the clustering results, and each node represents a clustering branch. The mean and variance of the energy efficiency index in each branch are calculated. Based on the mean and variance, branches can be pruned (i.e., branches with noise and excessive dispersion are removed) to form a tree structure. The branches of the tree structure represent steady-state operating modes (small fluctuations, low variance) and unsteady-state operating modes (large fluctuations, high variance). For example, if the mean of the energy efficiency index in a cluster branch is 10 and the variance is 0.5, it can be considered that the branch corresponds to a steady-state mode; while the mean of another branch is 15 and the variance is 4, it corresponds to an unsteady-state mode. These branches are displayed in a tree structure to form an actuator operating mode classification tree.

[0044] Specifically, the energy consumption source analysis is performed on the non-steady-state operation mode to generate the energy efficiency degradation characteristic map, including: Identify the operating state range of the non-steady-state operation mode and determine the threshold range of the energy efficiency influencing factor; Specifically, according to the operating mode classification tree, the operating state intervals representing non-steady-state operation (such as large vibration and obvious fluctuation) are identified. In these intervals, the distribution of energy efficiency influencing factors is statistically analyzed to determine a threshold range, which is used to distinguish normal energy consumption from abnormal energy consumption. For example, assuming that in the non-steady-state mode area, the statistical values ​​of energy efficiency influencing factors are mainly distributed between 15 and 25, the threshold range can be determined to be 15 to 25.

[0045] According to the threshold range of energy efficiency influencing factors, filter out instruction segments with abnormally increased energy consumption, extract abnormal pulse patterns related to the control instruction sequence, and mark the start and end times of the abnormal pulse patterns; Specifically, within the energy efficiency impact factor threshold range, the energy consumption data corresponding to the control instructions in different time periods are compared to screen out the instruction segments with sudden energy consumption increases, which usually appear in the form of pulses. By marking the start and end times, an abnormal pulse pattern is formed. For example, if the energy efficiency impact factor suddenly rises from a normal value of 10 to more than 20 within a certain period of time, and maintains for 50 milliseconds before quickly recovering, this period of time can be marked as an abnormal pulse pattern, and its start and end times are recorded respectively.

[0046] Taking the start and end time of the abnormal pulse mode as a reference, calculate the abnormal change amount of the electric actuator motor current signal and the hysteresis response amount of the valve position feedback signal; Specifically, according to the start and end times of the marked abnormal pulse mode, the motor current signal and the valve position feedback signal in the corresponding period are extracted and calculated: Abnormal variation: Indicates the difference between the motor current signal and the normal level during abnormal pulse mode.

[0047] Hysteresis response: It indicates the response delay of the valve position feedback signal to the control command, that is, the time difference between the signal change and the issuance of the command.

[0048] Based on the numerical correlation between the abnormal change of the motor current signal and the hysteresis response of the valve position feedback signal, an energy efficiency degradation characteristic spectrum is generated; specifically, the correlation coefficient between the abnormal change and the hysteresis response is used to analyze the correlation between the two, and a spectrum is constructed to reflect the energy efficiency degradation under the non-steady-state operation mode. The relationship between the two parameters can be represented by scatter plots, line graphs, etc. If the two are positively correlated, it means that there is a strong coupling between the energy consumption abnormality and the response lag.

[0049] Specifically, the energy efficiency degradation feature map is input into the pre-trained energy consumption optimization decision model, and a dynamic control strategy set is output, including: A training set of energy efficiency degradation feature maps of electric actuators is constructed based on historical operation data, and an energy consumption optimization decision model is trained through supervised learning. Specifically, historical operation data is used to generate energy efficiency degradation feature maps, and the maps are used as training samples to construct a training set.

[0050] The supervised learning method is used to train the energy consumption optimization decision model so that it can identify abnormal energy consumption based on the input degradation feature map.

[0051] During training, input data is used together with pre-labeled energy-efficiency mode labels for model learning.

[0052] Exemplarily, data from the past month is collected to generate several energy efficiency degradation maps, and it is marked which time periods are abnormal energy consumption patterns and which are normal patterns. These data are used to train a neural network model or a support vector machine model.

[0053] Normalizing the characteristic parameters of the energy efficiency degradation characteristic map to obtain a standardized characteristic input matrix; The standardized feature input matrix is ​​input into the pre-trained energy consumption optimization decision model to perform feature matching and energy efficiency pattern recognition; Specifically, the standardized feature input matrix is ​​used as the input of the energy consumption optimization decision model. After training, the model can automatically match the relationship between the input features and the historical energy efficiency patterns and identify the current energy efficiency status.

[0054] The model may use algorithms such as deep neural networks and random forests to automatically output prediction results based on the input data.

[0055] Exemplarily, after the normalized feature vector is input into the model, the model output result may be steady-state operation or non-steady-state operation, and corresponding control suggestions are given.

[0056] According to the feature matching and energy efficiency pattern recognition results, a dynamic control strategy set corresponding to the operating mode of the electric actuator is output; Specifically, the energy consumption optimization decision model generates a dynamic control strategy for the electric actuator based on the identified energy efficiency mode, including adjusting the motor current, changing the valve position, or modifying the control instruction frequency.

[0057] The output dynamic control strategy set is the control parameter combination that best matches the current operating mode.

[0058] For example, if the model identifies that the current operating state is non-steady-state, the output strategy may be to reduce the current setting value by 0.5 amps, extend the valve response time by 10 milliseconds, etc.

[0059] Specifically, after the dynamic control strategy set takes effect, by comparing the deviation between the energy efficiency transition trajectory before and after the implementation of the strategy and the theoretical optimal energy consumption curve, the weight distribution of the energy consumption optimization decision model is iteratively updated, and the final optimization control parameter package is output, including: Real-time record of the change trend of the motor current signal, valve position feedback signal and vibration spectrum data of the electric actuator, and draw the actual energy efficiency transition trajectory; Specifically, after the dynamic control strategy takes effect, the system continuously collects key operating data of the actuator and draws a trajectory diagram reflecting the changes in energy efficiency based on these data.

[0060] The trajectory diagram reflects the changing trend of energy consumption over time and can show the changes before and after the implementation of the control strategy.

[0061] For example, before regulation, the energy efficiency transition trajectory may show large fluctuations; after regulation, the trajectory tends to be stable and closer to the theoretical optimal curve.

[0062] Compare the actual energy efficiency transition trajectory with the theoretical optimal energy consumption curve point by point, and calculate the degree of deviation between the actual energy efficiency transition trajectory and the theoretical optimal energy consumption curve; Specifically, the actual collected energy efficiency data is compared with the preset theoretical optimal energy consumption curve, and the difference between the two at each time point is calculated to obtain the overall deviation degree. For example, if at a certain time point, the actual energy efficiency value is 8 and the theoretical value is 7, then the error at this point is 8-7=1; after calculating all time points, the mean square error value is taken as the overall deviation degree.

[0063] The deviation degree is fed back to the energy consumption optimization decision model as an error, and the weight distribution within the energy consumption optimization decision is iteratively adjusted based on the error back propagation mechanism; Specifically, an error back-propagation algorithm is used, and the calculated degree of deviation (such as the mean square error value) is used as the output of the loss function and fed back to the pre-trained energy consumption optimization decision model, thereby adjusting the weights of each layer within the model to make the next prediction more accurate.

[0064] When the degree of deviation falls within the preset allowable range, the final optimized control parameter package is output; Specifically, when, after iterative adjustment, the degree of deviation between the actual energy efficiency transition trajectory and the theoretical optimal energy consumption curve is lower than a preset allowable threshold (for example, the mean square error is lower than 0.1), the model is considered to have reached the optimal state. At this time, the final control parameter package is output to guide the subsequent operation mode switching of the electric actuator.

[0065] For example, if the preset allowable mean square error threshold is 0.1, when the model is iterated and the mean square error is measured to be 0.08, the final optimized control parameter package is output, which includes the adjusted motor current setting value, valve action time and other parameters.

[0066] Embodiment 2:

[0067] The difference between Example 2 of the present invention and Example 1 is that this example introduces an energy consumption optimization system for electric actuators based on big data analysis.

[0068] Figure 2 The present invention provides a schematic diagram of the structure of the electric actuator energy consumption optimization system based on big data analysis, which includes a feature extraction module, an influencing factor construction module, a working condition clustering module, an energy consumption tracing analysis module, a control strategy output module, and a model update module. The feature extraction module collects multi-source operation data streams of the electric actuator, extracts time series features from the multi-source operation data streams based on a sliding window, and generates a modal feature vector of the actuator operating condition; The influence factor construction module constructs an energy efficiency influence factor set based on the actuator working condition modal eigenvector and the preset energy efficiency influence factor weight matrix; The working condition clustering module clusters the working condition of the energy efficiency influencing factor set based on the dynamic density clustering algorithm and generates an actuator operation mode classification tree; The energy consumption source analysis module performs energy consumption source analysis on non-steady-state operation modes and generates energy efficiency degradation characteristic maps; The control strategy output module inputs the energy efficiency degradation feature map into the pre-trained energy consumption optimization decision model and outputs a dynamic control strategy set; After the dynamic control strategy set takes effect, the model update module iteratively updates the weight distribution of the energy consumption optimization decision model by comparing the deviation between the energy efficiency transition trajectory before and after the implementation of the strategy and the theoretical optimal energy consumption curve, and outputs the final optimization control parameter package.

[0069] The above formulas are all dimensionless and numerical calculations. The formula is a formula for the most recent real situation obtained by collecting a large amount of data and performing software simulation. The preset parameters and thresholds in the formula are set by technicians in this field according to actual conditions.

[0070] The above embodiments may be implemented in whole or in part by software, hardware, firmware or any other combination thereof. When implemented by software, the above embodiments may be implemented in whole or in part in the form of a computer program product. The computer program product includes one or more computer instructions or computer programs. When the computer instructions or computer programs are loaded or executed on a computer, the process or function described in the embodiment of the present application is generated in whole or in part. The computer may be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions may be stored in a computer-readable storage medium, or may be transmitted from one computer-readable storage medium to another computer-readable storage medium. For example, the computer instructions may be transmitted from one website, computer, server or data center to another website, computer, server or data center by wired (e.g., infrared, wireless, microwave, etc.) means. The computer-readable storage medium may be any available medium that a computer can access or a data storage device such as a server or data center that contains one or more available media sets. The available medium may be a magnetic medium (e.g., a floppy disk, a hard disk, a tape), an optical medium (e.g., a DVD), or a semiconductor medium. The semiconductor medium may be a solid-state hard disk.

[0071] Those of ordinary skill in the art will appreciate that the modules and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of this application.

[0072] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working processes of the systems, devices and modules described above can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0073] In the several embodiments provided in the present application, it should be understood that the disclosed systems, devices and methods can be implemented in other ways. For example, the device embodiments described above are only schematic. For example, the division of the modules is only a logical function division. There may be other division methods in actual implementation, such as multiple modules or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of devices or modules, which can be electrical, mechanical or other forms.

[0074] The modules described as separate components may or may not be physically separated, and the components shown as modules may or may not be physical modules, and may be located in one place or distributed on multiple network modules. Some or all of the modules may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0075] In addition, each functional module in each embodiment of the present application may be integrated into one processing module, or each module may exist physically separately, or two or more modules may be integrated into one module.

[0076] If the functions are implemented in the form of software function modules and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application can be essentially or partly embodied in the form of a software product that contributes to the prior art. The computer software product is stored in a storage medium, including several instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the methods described in the various embodiments of the present application. The aforementioned storage media include: U disk, mobile hard disk, read-only memory (ROM), random access memory (RAM), disk or optical disk, and other media that can store program codes.

[0077] The above is only a specific implementation of the present application, but the protection scope of the present application is not limited thereto. Any person skilled in the art who is familiar with the present technical field can easily think of changes or substitutions within the technical scope disclosed in the present application, which should be included in the protection scope of the present application. Therefore, the protection scope of the present application should be based on the protection scope of the claims.

[0078] Finally: The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the protection scope of the present invention.

Claims

1. An electric actuator energy consumption optimization method based on big data analysis, characterized in that: The steps include: Collect multi-source operation data streams of electric actuators, extract time series features of the multi-source operation data streams based on sliding windows, and generate actuator working condition modal feature vectors; According to the actuator working condition modal eigenvector and the preset energy efficiency influencing factor weight matrix, an energy efficiency influencing factor set is constructed; Based on the dynamic density clustering algorithm, the energy efficiency influencing factor set is clustered into working conditions to generate the actuator operation mode classification tree; Conduct energy consumption source analysis on non-steady-state operation modes and generate energy efficiency degradation characteristic maps; Input the energy efficiency degradation feature map into the pre-trained energy consumption optimization decision model and output a dynamic control strategy set; After the dynamic control strategy set takes effect, the weight distribution of the energy consumption optimization decision model is iteratively updated by comparing the deviation between the energy efficiency transition trajectory before and after the implementation of the strategy and the theoretical optimal energy consumption curve, and the final optimization control parameter package is output.

2. The method for optimizing energy consumption of electric actuators based on big data analysis according to claim 1, characterized in that: Collect the multi-source operation data stream of the electric actuator, extract the time series features of the multi-source operation data stream based on the sliding window, and generate the actuator working condition modal feature vector, specifically: Collect the motor current signal, valve position feedback signal, control instruction sequence and vibration spectrum data of the electric actuator to form a multi-source heterogeneous operation data stream; Adjust the size of the sliding window according to the frequency of changes in the control instruction sequence; Extract time series features from multi-source heterogeneous operation data streams within a sliding window; The time domain features, frequency domain features and wavelet transform features extracted in each sliding window are dimensionally spliced ​​to generate the actuator working condition modal feature vector.

3. The method for optimizing energy consumption of electric actuators based on big data analysis according to claim 2 is characterized in that: According to the actuator working condition modal eigenvector and the preset energy efficiency influencing factor weight matrix, the energy efficiency influencing factor set is constructed, which is as follows: Extracting time domain characteristic parameters, frequency domain characteristic parameters and wavelet transform characteristic parameters from the actuator working mode characteristic vector; Match the row vectors of the energy efficiency impact factor weight matrix with the characteristic parameters of the corresponding type respectively; Performing a weighted calculation operation on each feature parameter; The weighted calculation results are accumulated according to the feature type to generate a multi-dimensional energy efficiency influencing factor set.

4. The method for optimizing energy consumption of electric actuators based on big data analysis according to claim 3 is characterized in that: Based on the dynamic density clustering algorithm, the energy efficiency influencing factor set is clustered into working conditions to generate the actuator operation mode classification tree, which is as follows: The factor whose product of local density and relative distance in the multi-dimensional energy efficiency influencing factor concentration exceeds the threshold is selected as the initial cluster center; According to the initial cluster center distribution density, the continuous operating conditions with similar density and continuous distribution of energy efficiency influencing factors are divided into the same cluster branch; The discrete factors that meet the density connectivity conditions are classified into adjacent cluster branches; A tree classification structure is constructed, and pruning optimization is performed based on the mean and variance of the energy efficiency index of the clustering branches to generate an actuator operation mode classification tree including steady-state operation modes and non-steady-state operation modes.

5. The method for optimizing energy consumption of electric actuators based on big data analysis according to claim 4 is characterized in that: The energy consumption source analysis is performed on the non-steady-state operation mode to generate the energy efficiency degradation characteristic map, specifically: Identify the operating state range of the non-steady-state operation mode and determine the threshold range of the energy efficiency influencing factor; According to the threshold range of energy efficiency influencing factors, filter out instruction segments with abnormally increased energy consumption, extract abnormal pulse patterns related to the control instruction sequence, and mark the start and end times of the abnormal pulse patterns; Taking the start and end time of the abnormal pulse mode as a reference, calculate the abnormal change amount of the electric actuator motor current signal and the hysteresis response amount of the valve position feedback signal; Based on the numerical correlation between the abnormal change of the motor current signal and the hysteresis response of the valve position feedback signal, an energy efficiency degradation characteristic map is generated.

6. The method for optimizing energy consumption of electric actuators based on big data analysis according to claim 5, characterized in that: The energy efficiency degradation feature map is input into the pre-trained energy consumption optimization decision model, and a dynamic control strategy set is output, which is as follows: A training set of energy efficiency degradation feature maps of electric actuators is constructed based on historical operation data, and an energy consumption optimization decision model is trained through supervised learning. Normalizing the characteristic parameters of the energy efficiency degradation characteristic map to obtain a standardized characteristic input matrix; The standardized feature input matrix is ​​input into the pre-trained energy consumption optimization decision model to perform feature matching and energy efficiency pattern recognition; According to the feature matching and energy efficiency pattern recognition results, a dynamic control strategy set corresponding to the operating mode of the electric actuator is output.

7. The method for optimizing energy consumption of electric actuators based on big data analysis according to claim 6, characterized in that: After the dynamic control strategy set takes effect, the weight distribution of the energy consumption optimization decision model is iteratively updated by comparing the deviation between the energy efficiency transition trajectory before and after the implementation of the strategy and the theoretical optimal energy consumption curve, and the final optimization control parameter package is output, which is as follows: Real-time record of the change trend of the motor current signal, valve position feedback signal and vibration spectrum data of the electric actuator, and draw the actual energy efficiency transition trajectory; Compare the actual energy efficiency transition trajectory with the theoretical optimal energy consumption curve point by point, and calculate the degree of deviation between the actual energy efficiency transition trajectory and the theoretical optimal energy consumption curve; The deviation degree is fed back to the energy consumption optimization decision model as an error, and the weight distribution within the energy consumption optimization decision is iteratively adjusted based on the error back propagation mechanism; When the degree of deviation drops to within the preset allowable range, the optimized final control parameter package is output.

8. An electric actuator energy consumption optimization system based on big data analysis, used to implement the electric actuator energy consumption optimization method based on big data analysis according to any one of claims 1 to 7, characterized in that: It includes feature extraction module, influencing factor construction module, operating state clustering module, energy consumption source analysis module, control strategy output module and model update module; The feature extraction module collects multi-source operation data streams of the electric actuator, extracts time series features from the multi-source operation data streams based on a sliding window, and generates a modal feature vector of the actuator operating condition; The influence factor construction module constructs an energy efficiency influence factor set based on the actuator working condition modal eigenvector and the preset energy efficiency influence factor weight matrix; The working condition clustering module clusters the working condition of the energy efficiency influencing factor set based on the dynamic density clustering algorithm and generates an actuator operation mode classification tree; The energy consumption source analysis module performs energy consumption source analysis on non-steady-state operation modes and generates energy efficiency degradation characteristic maps; The control strategy output module inputs the energy efficiency degradation feature map into the pre-trained energy consumption optimization decision model and outputs a dynamic control strategy set; After the dynamic control strategy set takes effect, the model update module iteratively updates the weight distribution of the energy consumption optimization decision model by comparing the deviation between the energy efficiency transition trajectory before and after the implementation of the strategy and the theoretical optimal energy consumption curve, and outputs the final optimization control parameter package.

Images