Multi-modal error dynamic compensation method and system of intelligent electric meter

Abstract

The invention discloses a multi-modal error dynamic compensation method and system for an intelligent electric meter, and the method comprises the steps: constructing an environment perception unit, and obtaining an environment factor parameter; performing weighted calculation on the environmental factor parameters to generate an environmental metabolic index; constructing an error prediction model based on the hormone release promoting factor action mechanism in combination with the environmental metabolic index so as to obtain a prediction error offset; original signals are collected in real time, multi-modal error detection is carried out, and error vectors are generated; acquiring a compensation trigger threshold according to the error vector; and comparing the predicted error offset with a compensation trigger threshold, and triggering a multi-stage compensation mechanism. According to the invention, the metering parameters of the electric meter can be dynamically adjusted on line in real time under complex working conditions, so that the stability and accuracy of the electric meter are improved.

Description

Technical Field

[0001] The present invention relates to the technical field of smart electric meters, and in particular to a multi-modal error dynamic compensation method and system for smart electric meters. Background Art

[0002] With the rapid growth of global energy demand and the in-depth promotion of smart grids, smart meters, as the core equipment of the power metering system, are directly related to the fairness of energy settlement and the efficiency of power grid operation in terms of accuracy and reliability. However, in actual applications, smart meters are exposed to complex dynamic environments and operating condition fluctuations for a long time. Complex dynamic environments such as temperature drift, humidity changes, and electromagnetic interference, and operating condition fluctuations such as load mutations and device aging all lead to a significant increase in metering errors.

[0003] Traditional error compensation technologies mostly rely on static models or single parameter adjustments, and it is difficult to respond to multi-modal errors in real time, such as the synergistic effects of steady-state, transient, and coupling errors. In particular, compensation lag or over-adjustment problems are prone to occur under sudden interference, which seriously restricts the long-term stability and scenario adaptability of the electricity meter.

[0004] In the prior art, the invention patent with publication number CN118940146A discloses a temperature-sensing-based smart meter error correction method, medium, and electronic device. In the method, temperature data is collected in real time through thermocouples, and error correction is performed using a Bayesian anomaly detection algorithm and a multivariate Gaussian mixture model. This method performs well in error compensation caused by temperature changes, but fails to effectively solve error problems in complex environments such as electromagnetic interference and load fluctuations.

[0005] The invention patent with publication number CN118465337A discloses an error compensation method, device and electric energy meter for an electric energy meter. In the method, an excitation signal is injected into the sampling element and a frequency domain analysis is performed using discrete Fourier transform to achieve autonomous monitoring and compensation of errors. However, there are still limitations when dealing with the superposition effect of multimodal errors such as temperature, humidity and electromagnetic interference.

[0006] The invention patent with publication number CN119247256A discloses a full-temperature automatic calibration and compensation method for electric energy meters. The patent adopts a multi-stage non-uniform temperature segmentation point design to adapt to various temperature-error curves. This method has made a breakthrough in the automation of temperature compensation, but does not involve comprehensive compensation of other environmental factors such as humidity and electromagnetic interference. Summary of the invention

[0007] The technical problem to be solved by the present invention is to improve the measurement accuracy of smart meters under the complex dynamic environment and operating condition fluctuations of the power grid.

[0008] In order to solve the above technical problems, the present invention provides the following technical solutions:

[0009] A multi-modal error dynamic compensation method for a smart meter, comprising:

[0010] Construct an environmental perception unit to obtain environmental factor parameters;

[0011] Perform weighted calculation on environmental factor parameters to generate environmental metabolic index;

[0012] Based on the mechanism of action of hormone-releasing factor and combined with environmental metabolic index, an error prediction model is constructed to obtain the prediction error offset;

[0013] Collect original signals in real time, perform multi-modal error detection, and generate error vectors;

[0014] According to the error vector, a compensation trigger threshold is obtained;

[0015] The prediction error offset is compared with the compensation trigger threshold to trigger the multi-level compensation mechanism.

[0016] In one embodiment of the present invention, obtaining the environmental metabolic index includes:

[0017] A hormone-circuit equivalent model was established, and differential calculations were performed on the hormone-circuit equivalent model to obtain the weight coefficients of various environmental factor parameters and the environmental metabolic index.

[0018] In one embodiment of the present invention, the hormone-circuit equivalent model is:

[0019]

[0020] The hormone-circuit equivalent model is differentiated and calculated as:

[0021]

[0022] In the formula, C i is the compensation coefficient of the i-th error type, α, β, γ are the weight coefficients of the environmental factor parameters temperature, humidity and electromagnetic respectively, η i is the circuit transfer gain, V max , K m , n are enzyme kinetic parameters, V max is the maximum compensation capability, corresponding to the maximum gain of the compensation branch in the circuit, K m is the half-saturation coefficient, which indicates the [H] concentration required to achieve the maximum compensation effect and reflects the sensitivity of the compensation circuit. n is the Hill coefficient, [H] is mapped to the dynamic regulation signal intensity in the circuit. δ is the metabolic rate constant and d is the differential sign.

[0023] In one embodiment of the present invention, the error prediction model is:

[0024]

[0025] Where Δε is the prediction error offset, k1 and k2 are metabolic regulation coefficients, Environmental Metabolic Index I x The derivative of is the accumulation amount, environmental metabolic index I x The integral of , t is the time, and I x =αT+βH+γE M , T,H,E M are the environmental factor parameters temperature, humidity and electromagnetic, respectively; α, β, γ are the weight coefficients of the environmental factor parameters temperature, humidity and electromagnetic, respectively.

[0026] In one embodiment of the present invention, generating an error vector includes: performing discrete wavelet transform decomposition on an original signal to obtain an error vector; wherein the original signal is decomposed into:

[0027]

[0028] In the formula, S raw (t) is the original signal collected at time t, is the scale function, is the wavelet function, j is the decomposition level, g is the signal level, Denotes the approximate coefficient of the j0th layer when the signal level is g, D j,g It represents the detail coefficient of the jth layer of the decomposition level of the signal level g.

[0029] In one embodiment of the present invention, the approximation coefficient of the deepest decomposition level is taken as the steady-state error ε s , take the detail coefficient of the shallowest decomposition level as the transient error ε t , the coupling error ε is obtained by multi-scale phase analysis or energy cross calculation of the detail coefficients c , the obtained steady-state error, transient error and coupling error are formed into a vector to obtain the error vector Where T is the matrix transpose.

[0030] In one embodiment of the present invention, the compensation triggering threshold is obtained by the following formula:

[0031]

[0032] Where θ1 and θ2 are the compensation trigger thresholds of the multi-level compensation mode, ω s ,ω t are the error weight coefficients corresponding to the compensation trigger thresholds θ1 and θ2, ε s ,ε t ,ε cThey are steady-state error, transient error and coupling error, is the smoothing factor, θ 01 ,θ 02 are the initial values ​​of the compensation trigger thresholds θ1 and θ2 respectively.

[0033] In one embodiment of the present invention, the prediction error offset is compared with the compensation triggering threshold to trigger a multi-level compensation mechanism, including:

[0034] When Δε<θ1, the slow adjustment mode is executed, and the voltage reference value V ref , the compensation formula is: Among them, τ s is the time constant, C s is the steady-state compensation coefficient, is the voltage reference value after compensation, t′ is the accumulated time after compensation starts;

[0035] When θ1≤Δε<θ2, the fast response mode is executed and the current sampling rate f is dynamically adjusted. s , the compensation formula is: Where λ is the response sensitivity coefficient, is the current sampling rate after compensation;

[0036] When Δε≥θ2, the antagonistic balance mode is executed to generate the phase compensation Δφ. The compensation formula is: Among them, ε s ,ε t ,ε c are the steady-state error, transient error and coupling error respectively, and Δε is the prediction error offset, θ1 and θ2 are the compensation trigger thresholds of the multi-level compensation mode respectively.

[0037] In one embodiment of the present invention, the multi-level compensation mechanism is regularly updated to optimize the multi-level compensation mechanism:

[0038] After the multi-level compensation mechanism is triggered, the signal corrected by the compensation circuit is collected Calculate the residual error And feed back to the error prediction model to update the metabolic regulation coefficients k1 and k2. in To set the error threshold, is the updated Xie adjustment coefficient, S raw is the original signal, ||·|| is the norm operator;

[0039] Online optimization of enzyme kinetic parameters, introduction of compensation intensity periodic adjustment function Among them, A0 is the reference compensation amount, φ is the phase shift angle;

[0040] Based on historical error data, the enzyme kinetic parameter V max , K m Calculate partial derivatives, use gradient descent method to optimize parameters, and feed back to the hormone-circuit equivalent model to update the error type compensation coefficient C i .

[0041] The present invention also provides a multi-modal error dynamic compensation system for a smart meter, which uses the multi-modal error dynamic compensation method for a smart meter as described above, and comprises:

[0042] The hypothalamus hierarchical module is used to construct an environmental perception unit and obtain environmental factor parameters; perform weighted calculations on environmental factor parameters to generate an environmental metabolic index;

[0043] The pituitary level module is used to construct an error prediction model based on the mechanism of action of hormone-releasing factor and combined with the environmental metabolic index to obtain the prediction error offset;

[0044] A multi-modal error detection module is used to collect original signals in real time, perform multi-modal error detection, and generate error vectors;

[0045] The target gland level module is used to obtain the compensation trigger threshold according to the error vector; compare the predicted error offset with the compensation trigger threshold to trigger the multi-level compensation mechanism.

[0046] Compared with the prior art, the beneficial effects of the present invention are as follows: the design idea of ​​the present invention is to simulate the feedback regulation mechanism in the biological endocrine system to solve the error self-correction problem in the smart meter. The system includes an environmental perception module, an error prediction model and a compensation execution unit, and realizes real-time dynamic compensation through a three-level feedback mechanism (hypothalamus level, pituitary level, target gland level). It can adjust the metering parameters of the meter in real time and dynamically under complex working conditions, thereby improving the stability and accuracy of the meter. It mainly solves the following technical problems:

[0047] 1. The problem of dynamic coordinated correction of multi-modal errors (steady-state / transient / coupled errors)

[0048] 2. Real-time adaptive adjustment of compensation parameters under complex environmental interference;

[0049] 3. The defect of traditional compensation algorithm is that it cannot take into account both response speed and accuracy.

[0050] The present invention collects environmental parameters in real time through temperature, humidity, and electromagnetic sensors at the hypothalamus level (environmental perception unit) to construct an environmental metabolic index. This index quantifies the comprehensive impact of environmental interference on the metering chip and simulates the ability of an organism to perceive external stimuli. In high temperature and high humidity scenarios, the system automatically increases the temperature weight and prioritizes suppressing thermal drift errors; in strong electromagnetic interference scenarios, the electromagnetic coefficient is increased to enhance anti-interference capabilities.

[0051] Pituitary level (error prediction model): Based on the dynamic characteristics of hormone-releasing factor, a differential-integral hybrid prediction model is established. The model predicts the error offset in the future time window by analyzing the temporal change trend and cumulative effect of the environmental metabolic index. The steady-state error prediction focuses on the role of the integral term to capture long-term environmental changes; the transient error prediction strengthens the weight of the derivative term to quickly respond to sudden interference.

[0052] Target gland level (compensation execution unit): converts the predicted error offset into three types of compensation instructions:

[0053] 1. Slow adjustment of thyroid hormone: through gradual voltage reference adjustment, slow-changing errors such as temperature drift are suppressed;

[0054] 2. Adrenaline fast response: dynamically increase the current sampling rate to deal with transient interference such as surge impact;

[0055] 3. Insulin-glucagon antagonism: offsetting multimodal coupling errors through phase compensation.

[0056] The present invention classifies errors and triggers them according to priority. The system decomposes the original signal into three modes: steady state, transient state, and coupling through wavelet transform:

[0057] Steady-state error: triggers the thyroid hormone slow adjustment mode and corrects the voltage reference step by step;

[0058] Transient error: Activate the adrenaline fast response mode and increase the sampling rate instantaneously;

[0059] Coupling error: Invoke the antagonistic balance mode to dynamically calculate the phase offset.

[0060] Rhythm adaptive calibration, the present invention simulates the circadian rhythm characteristics of glucocorticoids and establishes a periodic adjustment function for compensation intensity. The system automatically resets the baseline parameters every 24 hours, optimizes the enzyme kinetic model in combination with historical error data, and achieves long-term stability. The daytime mode enhances compensation sensitivity to adapt to grid load fluctuations, while the night mode reduces compensation intensity and prioritizes energy-saving operation.

[0061] The multi-level bionic mechanism significantly reduces the steady-state error, and the pulse compensation strategy breaks through the transient response bottleneck and improves accuracy. Intermittent compensation and mode priority arbitration reduce invalid power consumption; the dual mechanism of environmental perception and rhythm calibration ensures stable operation under extreme conditions and strong robustness. The enzyme kinetic model supports online parameter updates, adapts to the personalized needs of different types of metering chips, and has good scalability. BRIEF DESCRIPTION OF THE DRAWINGS

[0062] Figure 1 The present invention is a flowchart of a method for dynamic compensation of multi-modal errors of a smart meter according to an embodiment of the present invention.

[0063] Figure 2 The present invention is a block diagram of a multi-modal error dynamic compensation system for a smart meter according to an embodiment of the present invention. DETAILED DESCRIPTION

[0064] In order to facilitate those skilled in the art to understand the technical solution of the present invention, the technical solution of the present invention is further described in conjunction with the accompanying drawings of the specification.

[0065] The terms "first" and "second" are used for descriptive purposes only and should not be understood as indicating or implying relative importance or implicitly indicating the number of the indicated technical features. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of the features. In the description of this application, the meaning of "plurality" is two or more, unless otherwise clearly and specifically defined.

[0066] Please participate Figure 1 As shown, the present invention provides a multi-modal error dynamic compensation method for a smart meter, comprising:

[0067] S10, constructing an environment perception unit to obtain environmental factor parameters.

[0068] In one embodiment of the present invention, by integrating temperature T, humidity H, electromagnetic E M The sensor array constructs an environmental perception unit to collect environmental parameters X={T,H,E M}.

[0069] S20, performing weighted calculation on environmental factor parameters to generate an environmental metabolic index.

[0070] In one embodiment of the present invention, the environmental metabolic index is 1 x =αT+βH+γE M , where T, H, E M are the environmental factor parameters temperature, humidity and electromagnetic, respectively; α, β, γ are the weight coefficients of the environmental factor parameters temperature, humidity and electromagnetic, respectively.

[0071] In this embodiment, the specific α, β, and γ values ​​are obtained by mapping the dynamic compensation algorithm parameters. Specifically, a mapping relationship between the hormone secretion kinetic equation and the compensation parameters is established, that is, a hormone-circuit equivalent model is established, and the hormone-circuit equivalent model is differentiated to obtain the weight coefficients of the environmental factor parameters and the environmental metabolic index.

[0072] In this embodiment, the hormone-circuit equivalent model is:

[0073]

[0074] In the formula, Ci is the compensation coefficient of the i-th error type, α, β, γ are the weight coefficients of the environmental factor parameters temperature, humidity and electromagnetic respectively, η i is the circuit transfer gain, V max , K m , n are enzyme kinetic parameters, V max is the maximum compensation capability, corresponding to the maximum gain of the compensation branch in the circuit, K m is the half-saturation coefficient, which indicates the [H] concentration required to achieve the maximum compensation effect and reflects the sensitivity of the compensation circuit. n is the Hill coefficient, and [H] is mapped to the dynamic adjustment signal intensity in the circuit.

[0075] In this embodiment, the error types defined include steady-state error, transient error and coupling error. Steady-state errors include long-term offsets such as temperature drift, transient errors such as transient interferences such as surge impact, and coupling errors such as the superposition effect of electromagnetic interference and mechanical vibration. And the compensation branch in the circuit is a voltage compensation branch, which specifically simulates the slow adjustment of thyroid hormone, gradually adjusts the voltage reference value, a current compensation branch, which specifically simulates the fast response of adrenaline, dynamically adjusts the current sampling rate, and a phase compensation branch, which specifically simulates the insulin-glucagon antagonistic balance and corrects the coupling error phase offset.

[0076] In this embodiment, the real-time calculation of the hormone-circuit differential equations is realized by hardware description language such as HDL:

[0077]

[0078] Among them, δ is the metabolic rate constant, , the compensation parameter adjusts the speed, and d is the differential symbol.

[0079] S30, based on the mechanism of action of troponin-releasing factor and combined with the environmental metabolic index, an error prediction model was constructed to obtain the prediction error offset.

[0080] In one embodiment of the present invention, the error prediction model is:

[0081]

[0082] Where Δε is the prediction error offset, k1 and k2 are metabolic regulation coefficients, Environmental Metabolic Index I x The derivative of is the accumulation amount, environmental metabolic index I x The integral of , t is the time.

[0083] S40, collecting original signals in real time, performing multi-modal error detection, and generating error vectors.

[0084] In one embodiment of the present invention, the original measurement signal S is collected by a high-precision ADC circuit module. raw , perform discrete wavelet transform DWT, decompose into multi-level approximation coefficients and detail coefficients, use Daubechies wavelet to decompose the signal through convolution and downsampling operations. Specifically, the original signal is decomposed into: In the formula, S raw (t) is the original signal collected at time t, is the scale function, is the wavelet function, j is the decomposition level, g is the signal level, Denotes the approximate coefficient of the j0th layer when the signal level is g, D j,g It represents the detail coefficient of the jth layer of the decomposition level of the signal level g.

[0085] In this embodiment, the approximate coefficient of the deepest decomposition level is taken as the steady-state error ε s , take the detail coefficient of the shallowest decomposition level as the transient error ε t , the coupling error ε is obtained by multi-scale phase analysis or energy cross calculation of the detail coefficients c , the obtained steady-state error, transient error and coupling error are formed into a vector to obtain the error vector Where T is the matrix transpose.

[0086] In this embodiment, specifically, the approximate coefficient a of the deepest decomposition level is taken 5,g is the steady-state error ε s , take the detail coefficient D of the shallowest decomposition level 1,g is the transient error ε t , coupling error ε c The calculation formula is

[0087] S50: Obtain a compensation trigger threshold according to the error vector.

[0088] In one embodiment of the present invention, the built-in sensor of the metering chip in the meter collects environmental factor parameters and calculates the environmental metabolic index. The data is transmitted to the FPGA chip preprocessing module through the SPI interface. The environmental metabolic index is input into the error prediction model, and the prediction error offset Δε is output to achieve the effect of multi-modal error synchronous detection and metering signal fusion. The high-speed ADC circuit module is used to collect the original electrical signal S in real time. raw , decomposed into steady-state, transient, and coupled components through wavelet transform, generating the error vector The threshold used to calculate Δε triggers different compensation mechanisms. The compensation trigger threshold is obtained by the following formula:

[0089]

[0090] Where θ1 and θ2 are the compensation trigger thresholds of the multi-level compensation mode, ω s ,ω t are the error weight coefficients corresponding to the compensation trigger thresholds θ1 and θ2, respectively, satisfying ω s +ω t =1,ε s ,ε t ,ε c They are steady-state error, transient error and coupling error, is the smoothing factor, θ 01 ,θ 02 are the initial values ​​of the compensation trigger thresholds θ1 and θ2 respectively.

[0091] S60, comparing the prediction error offset with the compensation triggering threshold, triggering a multi-level compensation mechanism.

[0092] In one embodiment of the present invention, the triggering of the multi-level compensation mechanism is based on the amplitude and change rate of the predicted error offset Δε, and Δε is compared with the compensation triggering threshold, and the compensation execution unit triggers the three-level compensation mechanism. ref Affects the overall accuracy of the system. Adjustment should avoid sudden changes. Use slow adjustment mode. The current sampling rate f s It is necessary to quickly track transient changes and adopt a fast response mode. The coupling error needs to be adjusted bidirectionally. The phase compensation Δφ balances the orthogonal components through the inverse tangent function, corresponding to the antagonistic balance mode. The details are as follows:

[0093] When Δε<θ1, the slow adjustment mode is executed, the thyroxine analog compensation is called, and the voltage reference value V is adjusted. ref , the compensation formula is: Among them, τ s is the time constant, C s is the steady-state compensation coefficient, is the voltage reference value after compensation, and t′ is the accumulated time after compensation is started.

[0094] When θ1≤Δε<θ2, the fast response mode is executed, adrenaline simulation compensation is activated, and the current sampling rate f is dynamically adjusted. s , the compensation formula is: Where λ is the response sensitivity coefficient, is the current sampling rate after compensation, and e is a natural constant.

[0095] When Δε≥θ2, the antagonistic balance mode is executed to generate the phase compensation Δφ. The compensation formula is: Among them, ε s ,ε t ,ε care steady-state error, transient error and coupling error respectively, arctan is the inverse tangent function, Δε is the prediction error offset, θ1 and θ2 are the compensation trigger thresholds of the multi-level compensation mode respectively.

[0096] In one embodiment of the present invention, the multi-modal error dynamic compensation method of a smart meter further includes the following steps: regularly updating the multi-stage compensation mechanism and optimizing the multi-stage compensation mechanism.

[0097] In this embodiment, three independent hardware threads allocated inside the FPGA chip are controlled in parallel. They are responsible for the gradual adjustment of the voltage reference, which is updated every 100ms, corresponding to the slow adjustment, managing the dynamic switching of the current sampling rate, with a response time of <10μs, corresponding to the fast adjustment, and performing the phase compensation calculation, which is refreshed every millisecond, corresponding to the antagonistic adjustment.

[0098] In this embodiment, hardware self-test and sensor calibration are performed. The FPGA chip starts the self-test program to verify the functional integrity of each module such as ADC, DAC, and compensation circuit. Baseline data is collected in an interference-free environment, and the environmental sensor zero point calibration is performed. The offset of each sensor is calculated and written to the non-volatile memory. The preset biological metabolism parameters such as V are read from the ROM. max , K m , n enzyme kinetic parameters, set the 24-hour cycle clock through the RTC module, initialize the rhythm regulation function A(t) (simulating glucocorticoid regulation), configure the FPGA chip timing interrupt, trigger data collection every millisecond, and perform parameter optimization once an hour.

[0099] In this embodiment, specifically, optimizing the multi-level compensation mechanism includes:

[0100] Optimize parameter dynamics and compensate parameters Δφ is written into the FPGA control register to drive the compensation circuit to perform correction. Collect the corrected signal Calculate the residual error And feed back to the error prediction model to update the weight coefficients k1 and k2: in To set the error threshold, is the updated Xie adjustment coefficient, S raw is the original signal, ||·|| is the norm operator.

[0101] Online optimization of enzyme kinetic parameters, introduction of compensation intensity periodic adjustment function Among them, A0 is the reference compensation amount, and φ is the phase offset angle. Rhythm adaptive calibration automatically calls the rhythm adjustment function A(t) every 24 hours to reset the compensation reference amount A0.

[0102] Based on historical error data, the enzyme kinetic parameter V max , K m Calculate partial derivatives, use gradient descent method to optimize parameters, and feed back to the core model of dynamic compensation algorithm, that is, hormone-circuit equivalent model, to update error type compensation coefficient C i .

[0103] See also Figure 2 As shown, the present invention also provides a multi-modal error dynamic compensation system for a smart meter, which uses the multi-modal error dynamic compensation method for a smart meter as described above, and comprises:

[0104] The hypothalamus hierarchical module is used to construct an environmental perception unit, obtain environmental factor parameters, perform weighted calculations on the environmental factor parameters, and generate an environmental metabolic index.

[0105] The pituitary level module is used to construct an error prediction model based on the mechanism of action of hormone-releasing factor combined with the environmental metabolic index to obtain the prediction error offset.

[0106] The multi-modal error detection module is used to collect the original signal in real time, perform multi-modal error detection, and generate an error vector.

[0107] The target gland level module is used to obtain the compensation trigger threshold according to the error vector; compare the predicted error offset with the compensation trigger threshold to trigger the multi-level compensation mechanism.

[0108] It is obvious to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the present invention can be implemented in other specific forms without departing from the spirit or essential features of the present invention. Therefore, the embodiments should be regarded as exemplary and non-limiting from any point of view, and the scope of the present invention is defined by the appended claims rather than the above description, and it is intended that all changes falling within the meaning and scope of the equivalent elements of the claims are included in the present invention, and any reference numerals in the claims should not be regarded as limiting the claims involved.

[0109] The above-described embodiments merely represent implementation methods of the invention. The protection scope of the present invention is not limited to the above-described embodiments. For those skilled in the art, several modifications and improvements may be made without departing from the concept of the present invention, which all belong to the protection scope of the present invention.

Claims

1. A multi-modal error dynamic compensation method for a smart meter, characterized in that: include: Construct an environmental perception unit to obtain environmental factor parameters; Perform weighted calculation on environmental factor parameters to generate environmental metabolic index; Based on the mechanism of action of hormone-releasing factor and combined with environmental metabolic index, an error prediction model is constructed to obtain the prediction error offset; Collect original signals in real time, perform multi-modal error detection, and generate error vectors; According to the error vector, a compensation trigger threshold is obtained; The prediction error offset is compared with the compensation trigger threshold to trigger the multi-level compensation mechanism.

2. The multi-modal error dynamic compensation method of a smart meter according to claim 1, characterized in that: Obtain environmental metabolic indices, including: A hormone-circuit equivalent model was established, and differential calculations were performed on the hormone-circuit equivalent model to obtain the weight coefficients of various environmental factor parameters and the environmental metabolic index.

3. The multi-modal error dynamic compensation method of the smart meter according to claim 2, characterized in that: The hormone-circuit equivalent model is: The hormone-circuit equivalent model is differentiated and calculated as: In the formula, C i is the compensation coefficient of the i-th error type, α, β, γ are the weight coefficients of the environmental factor parameters temperature, humidity and electromagnetic respectively, η i is the circuit transfer gain, V max , K m , n are enzyme kinetic parameters, V max is the maximum compensation capability, corresponding to the maximum gain of the compensation branch in the circuit, K m is the half-saturation coefficient, which indicates the [H] concentration required to achieve the maximum compensation effect and reflects the sensitivity of the compensation circuit. n is the Hill coefficient, [H] is mapped to the dynamic regulation signal intensity in the circuit. δ is the metabolic rate constant and d is the differential sign.

4. The multi-modal error dynamic compensation method of a smart meter according to claim 1, characterized in that: The error prediction model is: Where Δε is the prediction error offset, k1 and k2 are metabolic regulation coefficients, Environmental Metabolic Index I x The derivative of is the accumulation amount, environmental metabolic index I x The integral of , t is the time, and I x =αT+βH+γE M , T,H,E M are the environmental factor parameters temperature, humidity and electromagnetic, respectively; α, β, γ are the weight coefficients of the environmental factor parameters temperature, humidity and electromagnetic, respectively.

5. The multi-modal error dynamic compensation method of a smart meter according to claim 1, characterized in that: Generating an error vector includes: performing discrete wavelet transform decomposition on the original signal to obtain the error vector; wherein the original signal is decomposed into: In the formula, S raw (t) is the original signal collected at time t, is the scale function, is the wavelet function, j is the decomposition level, g is the signal level, Denotes the approximate coefficient of the j0th layer when the signal level is g, D j,g It represents the detail coefficient of the jth layer of the decomposition level of the signal level g.

6. The multi-modal error dynamic compensation method of a smart meter according to claim 5, characterized in that: Take the approximate coefficient of the deepest decomposition level as the steady-state error ε s , take the detail coefficient of the shallowest decomposition level as the transient error ε t , the coupling error ε is obtained by multi-scale phase analysis or energy cross calculation of the detail coefficients c , the obtained steady-state error, transient error and coupling error are formed into a vector to obtain the error vector Where T is the matrix transpose.

7. The multi-modal error dynamic compensation method of a smart meter according to claim 1, characterized in that: Get the compensation trigger threshold using the following formula: Where θ1 and θ2 are the compensation trigger thresholds of the multi-level compensation mode, ω s ,ω t are the error weight coefficients corresponding to the compensation trigger thresholds θ1 and θ2, ε s ,ε t ,ε c They are steady-state error, transient error and coupling error, is the smoothing factor, θ 01 ,θ 02 are the initial values ​​of the compensation trigger thresholds θ1 and θ2 respectively.

8. The multi-modal error dynamic compensation method of a smart meter according to claim 1, characterized in that: Comparing the prediction error offset with the compensation trigger threshold triggers a multi-level compensation mechanism, including: When Δε<θ1, the slow adjustment mode is executed, and the voltage reference value V ref , the compensation formula is: Among them, τ s is the time constant, C s is the steady-state compensation coefficient, is the voltage reference value after compensation, t ′ To compensate for the accumulated time after startup; When θ1≤Δε<θ2, the fast response mode is executed and the current sampling rate f is dynamically adjusted. s , the compensation formula is: Where λ is the response sensitivity coefficient, is the current sampling rate after compensation; When Δε≥θ2, the antagonistic balance mode is executed to generate the phase compensation Δφ. The compensation formula is: Among them, ε s ,ε t ,ε c are the steady-state error, transient error and coupling error respectively, and Δε is the prediction error offset, θ1 and θ2 are the compensation trigger thresholds of the multi-level compensation mode respectively.

9. The multi-modal error dynamic compensation method of a smart meter according to claim 6, characterized in that: Regularly update and optimize the multi-level compensation mechanism: After the multi-level compensation mechanism is triggered, the signal corrected by the compensation circuit is collected Calculate the residual error And feed back to the error prediction model to update the metabolic regulation coefficients k1 and k2. in To set the error threshold, is the updated Xie adjustment coefficient, S raw is the original signal, ||·|| is the norm operator; Online optimization of enzyme kinetic parameters, introduction of compensation intensity periodic adjustment function Among them, A0 is the reference compensation amount, φ is the phase shift angle; Based on historical error data, the enzyme kinetic parameter V max , K m Calculate partial derivatives, use gradient descent method to optimize parameters, and feed back to the hormone-circuit equivalent model to update the error type compensation coefficient C i .

10. A multi-modal error dynamic compensation system for a smart meter, characterized in that: The multi-modal error dynamic compensation method of the smart meter according to any one of claims 1 to 9 comprises: The hypothalamus hierarchical module is used to construct an environmental perception unit and obtain environmental factor parameters; perform weighted calculations on environmental factor parameters to generate an environmental metabolic index; The pituitary level module is used to construct an error prediction model based on the mechanism of action of hormone-releasing factor and combined with the environmental metabolic index to obtain the prediction error offset; A multi-modal error detection module is used to collect original signals in real time, perform multi-modal error detection, and generate error vectors; The target gland level module is used to obtain the compensation trigger threshold according to the error vector; compare the predicted error offset with the compensation trigger threshold to trigger the multi-level compensation mechanism.

Images