Intelligent agricultural machinery operation speed monitoring method and system based on multi-sensor information fusion

Abstract

The present invention discloses a method and system for intelligent agricultural machinery operation speed monitoring based on multi-sensor information fusion, wherein the method includes: obtaining three types of speed data: first speed data measured by an encoder speed measuring device, second speed data measured by a GPS speed measuring device, and third speed data measured by a Doppler radar speed measuring device; obtaining the agricultural machinery operation speed based on the three speed data, specifically including: processing each speed data using a weighted average filtering algorithm and fusing the processed three speed data to obtain a first predicted speed; fusing the three speed data using a Kalman filtering algorithm to obtain a second predicted speed; and fusing the first predicted speed and the second predicted speed using a Dependence-Scientific Evidence Theory algorithm, and using the fused speed prediction result as the agricultural machinery operation speed. The speed prediction result obtained by the data fusion method of the present invention is more accurate than the prediction result obtained by a single data fusion method.

Description

Technical Field

[0001] The present invention relates to the technical field of speed monitoring of agricultural machinery, and in particular to an intelligent agricultural machinery operation speed monitoring method and system based on multi-sensor information fusion. Background Art

[0002] In the development of modern agricultural machinery automation, the accuracy of speed measurement has a crucial impact on the quality of agricultural machinery operations. When operating in the field, agricultural machinery's travel and operating speeds need to be dynamically adjusted based on the terrain, soil conditions, and operational requirements to ensure that the machinery efficiently and accurately completes tasks such as seeding, tilling, and spraying. However, due to the complexity of the agricultural operating environment, such as variable ground resistance, humidity fluctuations, and slope differences between different plots of land, it is often difficult to accurately obtain and predict speed using only a single sensor or measurement method.

[0003] Existing technologies primarily rely on sensors such as encoders, GPS modules, and Doppler radar to measure the speed of agricultural machinery. These sensors are significantly affected by the environment. For example, when agricultural machinery is traveling on soft soil or slippery surfaces, the wheel speed sensor (encoder) may slip and become inaccurate. Furthermore, when the GPS signal is weak, the data generated by the GPS module may also be inaccurate or intermittent.

[0004] In the prior art, the electronic control system for a seed drill disclosed in patent CN 208590238 U determines whether to use ground wheel signals or GPS / Beidou signals based on the speed value. This processing method is relatively simple, but the resulting speed accuracy is still low. Patent CN118549672A discloses a method for measuring machine speed that uses weighted distribution to process ground wheel speed and positioning speed. Furthermore, in the field of robotics, data fusion methods such as Kalman filtering are used to process multiple sensor signals. These methods still occasionally result in inaccurate data, and there is still room for improvement in speed measurement stability. Summary of the Invention

[0005] Purpose of the invention: In order to overcome the deficiencies in the prior art, the present invention provides an intelligent agricultural machinery operation speed monitoring method and system based on multi-sensor information fusion, which can improve the accuracy and stability of machine speed detection.

[0006] Technical solution: To achieve the above-mentioned purpose, the present invention provides an intelligent agricultural machinery operation speed monitoring method based on multi-sensor information fusion, the method comprising:

[0007] Acquire three types of speed data: first speed data a1 measured by an encoder speed measuring device, second speed data a2 measured by a GPS speed measuring device, and third speed data a3 measured by a Doppler radar speed measuring device;

[0008] The operating speed of agricultural machinery is obtained based on the three speed data;

[0009] It is characterized in that: the agricultural machinery operating speed obtained based on the three speed data includes:

[0010] The speed data are processed using a weighted average filtering algorithm, and the three speed data are fused to obtain the first predicted speed V1;

[0011] The three speed data are fused using the Kalman filter algorithm to obtain the second predicted speed V2;

[0012] The first predicted speed V1 and the second predicted speed V2 are fused using the DS evidence theory algorithm, and the fused speed prediction result V is used as the operating speed of the agricultural machinery.

[0013] Furthermore, the weighted average filtering algorithm is used to process each speed data, and the three speed data are fused to obtain the first predicted speed V1, which includes:

[0014] Get the speed data a corresponding to each i (t) time series, respectively denoted as

[0015] A i ={a i (t1),a i (t2),…,a i (t n )};

[0016] Among them, t1, t2, …, t n is the time point; i=1,2,3;

[0017] Definition corresponds to the time series A i The weight of each element in ω(j);

[0018] Use the following formula to get the processed speed data corresponding to the current moment:

[0019]

[0020] Where c is the current time point, j is the index of the historical data point in the weighted moving sliding window;

[0021] The current first predicted speed V1 is obtained by weighted calculation based on the processed three speed data:

[0022]

[0023] Among them, α, β, and γ are weight coefficients.

[0024] Furthermore, corresponding to time series A i The weight ω(j) of each element in is specifically:

[0025]

[0026] Among them, ω(j) represents the value corresponding to a j is the weight of ; σ is the standard deviation.

[0027] Furthermore, the Kalman filter algorithm is used to fuse the three speed data to obtain the second predicted speed V2, including:

[0028] Define the state variable of the system as the current speed prediction value x k , the measurement vector is the three speed data z provided by the sensor k =[a1,a2,a3] T , state transition matrix A k =1, control variable matrix B k = 0, the conversion matrix H from the state vector to the measurement vector k =[1,1,1] T / 3, system interference covariance matrix Q, and measurement interference covariance matrix R;

[0029] Initialize the initial value of the state variable Initialize the state covariance matrix P0;

[0030] At each time step k, the velocity is first predicted according to the state transfer equation, and the prior state estimate at the current moment is calculated:

[0031]

[0032] And calculate the prior state covariance:

[0033]

[0034] The Kalman gain is calculated from the measurement equation:

[0035]

[0036] Use the Kalman gain to update the posterior state estimate of the current velocity:

[0037]

[0038] And update the posterior state covariance:

[0039]

[0040] in: and The posterior state estimates P at time k-1 and time k are k-1 and P k , respectively, the posterior estimated covariances at time k-1 and time k; is the prior estimated covariance at time k; K k is the Kalman gain;

[0041] The posterior state estimate for each time step As the second predicted speed V2 after fusion.

[0042] Furthermore, the method of fusing the first predicted speed V1 and the second predicted speed V2 using the DS evidence theory algorithm and using the fused speed prediction result V as the operating speed of the agricultural machinery includes:

[0043] Define the recognition framework Θ = {V1, V2} and construct the power set 2 Θ ={Φ,{V1},{V2},{V1,V2}};

[0044] Define basic trust assignment functions m1 and m2 for each subset within the identification framework Θ, where m1({V1}) and m2({V2}) represent the direct trust values ​​for V1 and V2 respectively, and m1({V1, V2}) and m2({V1, V2}) represent the trust values ​​that hold for both, reflecting the uncertainty part. The basic trust allocation function satisfies the following conditions:

[0045]

[0046] in, That is to say and m(A) is the basic trust allocation function corresponding to subset A, where A is a non-empty subset;

[0047] Calculate the conflict coefficient K to measure the degree of conflict between V1 and V2; the formula is:

[0048]

[0049] Among them: A i and B j Represent the focal elements of the basic trust allocation functions m1 and m2 respectively;

[0050] The Dempster combination rule is used to calculate the value of the fused basic trust distribution function m(A) for each non-empty subset A:

[0051]

[0052] According to the fused basic trust distribution function m(A), the final speed prediction result V is calculated.

[0053] An intelligent agricultural machinery operation speed monitoring system based on multi-sensor information fusion includes an encoder speed measuring device, a GPS speed measuring device and a Doppler radar speed measuring device, and also includes a controller connected to the three speed measuring devices. The controller can implement the above-mentioned intelligent agricultural machinery operation speed monitoring method based on multi-sensor information fusion.

[0054] Beneficial effects: The intelligent agricultural machinery operation speed monitoring method and system based on multi-sensor information fusion of the present invention has the following beneficial effects:

[0055] The present invention achieves precise monitoring of agricultural machinery operating speeds through multi-sensor information fusion, comprehensively utilizing weighted average filtering, Kalman filtering, and DS evidence theory algorithms to fully leverage the complementary advantages of different sensor data. Weighted average filtering improves the smoothness and noise resistance of the data, Kalman filtering enhances adaptability to dynamic changes and prediction accuracy, and DS evidence theory further integrates different prediction results, improving the reliability and accuracy of decision-making. The overall solution is highly robust and real-time, effectively coping with complex agricultural machinery operating environments and improving the accuracy and stability of operating speed monitoring. This agricultural machinery operating speed monitoring method and system is suitable for monitoring agricultural machinery operations such as sowing, spraying, and harvesting. BRIEF DESCRIPTION OF THE DRAWINGS

[0056] Figure 1 This is a flow chart of a method for monitoring the speed of agricultural machinery operation;

[0057] Figure 2 This is a schematic diagram of the specific process of obtaining the operating speed of agricultural machinery based on three types of speed data. DETAILED DESCRIPTION

[0058] The present invention will be further described below with reference to the accompanying drawings.

[0059] The intelligent agricultural machinery operation speed monitoring method based on multi-sensor information fusion of the present invention is based on an agricultural machinery operation speed monitoring system. The agricultural machinery operation speed monitoring system includes an encoder speed measuring device, a GPS speed measuring device and a Doppler radar speed measuring device, and also includes a controller connected to the three speed measuring devices. The controller can implement the following intelligent agricultural machinery operation speed monitoring method based on multi-sensor information fusion.

[0060] like Figure 1 As shown, the method includes the following steps S1-S2:

[0061] Step S1, obtaining three kinds of speed data: first speed data a1 measured by an encoder speed measuring device, second speed data a2 measured by a GPS speed measuring device, and third speed data a3 measured by a Doppler radar speed measuring device;

[0062] Step S2, obtaining the operating speed of the agricultural machinery based on the three speed data;

[0063] like Figure 2 As shown, the step S2 includes the following steps S201-S203:

[0064] Step S201, using a weighted average filtering algorithm to process each speed data, and fusing the processed three speed data to obtain a first predicted speed V1;

[0065] Step S202: Using the Kalman filter algorithm to fuse the three speed data to obtain a second predicted speed V2;

[0066] Step S203 , performing data fusion on the first predicted speed V1 and the second predicted speed V2 using the DS evidence theory algorithm, and using the fused speed prediction result V as the operating speed of the agricultural machinery.

[0067] This invention achieves precise monitoring of agricultural machinery operating speed through multi-sensor information fusion. It leverages the complementary advantages of different sensor data by combining weighted average filtering, Kalman filtering, and DS evidence theory. Weighted average filtering improves data smoothness and noise immunity, while Kalman filtering enhances adaptability to dynamic changes and predictive accuracy. DS evidence theory further integrates different prediction results, improving decision-making reliability and accuracy. The overall solution is highly robust and real-time, effectively adapting to complex agricultural machinery operating environments and enhancing the accuracy and stability of operating speed monitoring.

[0068] The process of processing the speed data using the weighted average filtering algorithm in step S201 and fusing the three speed data to obtain the first predicted speed V1 includes the following steps S301-S304:

[0069] Step S301, obtaining the speed data a corresponding to each i (t) time series, respectively denoted as

[0070] A i ={a i (t1),a i (t2),…,a i (t n )};

[0071] Among them, t1, t2, …, t n is the time point; i=1,2,3;

[0072] Step S302, define the time series A i The weight of each element in ω(j);

[0073] Step S303: Obtain the processed speed data corresponding to the current moment using the following formula:

[0074]

[0075] Where c is the current time point, j is the index of the historical data point in the weighted moving sliding window, and the sliding window only contains the current time point and the previous k data points;

[0076] Step S304: Perform weighted calculation based on the processed three speed data to obtain the current first predicted speed V1:

[0077]

[0078] Among them, α, β, and γ are weight coefficients used to indicate the importance of the speed data of each sensor.

[0079] The above step S302 corresponds to time series A i The weight ω(j) of each element in is specifically:

[0080]

[0081] Among them, ω(j) represents the value corresponding to a j is the weight of ; σ is the standard deviation.

[0082] This weighting function takes into account the distance between time point j and the current time point c. Data points closer in distance receive higher weights, while data points farther away receive lower weights. In practical applications, an appropriate standard deviation σ is often selected to control the width of the weighting function, thereby balancing the smoothing effect and the speed of response to data points. The weighted average filtering algorithm described above makes the calculated average more responsive to recent data and less responsive to more distant data, thus achieving a smoothing effect and avoiding the impact of abnormal data fluctuations.

[0083] This algorithm smoothes multi-sensor speed data using a weighted moving average filter. Using weighted distribution to prioritize recent data, it improves responsiveness to dynamic changes while mitigating noise interference. The improved algorithm relies solely on current and historical data, ensuring real-time performance. It is suitable for smoothing and predicting speed signals in dynamic scenarios, enhancing data fusion accuracy and system stability.

[0084] In the above step S304, in order to make the fusion speed more accurate, the weight coefficient needs to be dynamically adjusted, which specifically includes the following steps:

[0085] Step A1: Initialize weights α0, β0, and γ0 for the encoder, GPS, and Doppler radar based on the theoretical accuracy of the sensor and typical applicable scenarios. The initial weights satisfy the normalization condition: α0+β0+γ0=1.

[0086] Step A2: collect the speed data a1(t), a2(t), and a3(t) of each sensor, and use the sliding window to calculate the standard deviation σ of each sensor in the window. i :

[0087]

[0088] in: is the mean of the velocity data in the window, k is the window size, σ i The smaller the value, the higher the stability.

[0089] Step A3, based on Bayesian inference, combines historical measurement performance with current operating conditions to calculate the credibility P i :

[0090]

[0091] Where: E i Condition; P(a i |E i ) is the probability density of the speed measured in the current environment, P(E i ) is the prior probability of the environmental condition, P(a i ) is the marginal probability of the sensor velocity data;

[0092] Step A4: Calculate the suitability score S of each sensor using external auxiliary sensor data (such as humidity and terrain detection) i :

[0093] S i =f(H,T,C,P)

[0094] Function f is an empirical model for specific environmental variables or a suitability calculation formula obtained through training; H is humidity, T is terrain characteristics, C is temperature, and P is vibration intensity.

[0095] Step A5: Integrate the stability, credibility and applicability scores to calculate the comprehensive score W for each sensor. i :

[0096]

[0097] Among them, λ1, λ2, and λ3 are proportional factors for adjusting the contribution of the three weights, which are adjusted according to the actual scenario;

[0098] Step A6: Normalize the comprehensive score and update the weights α, β, and γ of each sensor:

[0099]

[0100] Step A7: Perform weighted fusion on the speed data using the updated weights to generate a first predicted speed V1 at the current moment.

[0101] This method achieves adaptive fusion of data from multiple sensors, significantly improving the accuracy, robustness, and real-time performance of speed monitoring. During agricultural machinery operation, the weight distribution of each sensor can be dynamically adjusted based on the real-time environment, avoiding global measurement errors caused by single sensor failure. Furthermore, by introducing the triple metrics of stability, reliability, and environmental suitability, the system's adaptability to complex environments is effectively enhanced, ensuring stable speed measurement output. This technology is particularly suitable for scenarios where agricultural machinery operates in variable terrain conditions, providing reliable technical support for improving the efficiency and safety of intelligent agricultural machinery.

[0102] In step A4 above, the specific calculation formula for the suitability score is:

[0103]

[0104] Among them, X j is the measured value of the jth environmental variable, such as current humidity or terrain parameters; Represents sensor i's response to environmental variable X j The optimal fitness value of , which is usually set based on experiments or historical data; is the adaptation range parameter, which defines the sensor i in the variable X j The nonlinear function part adopts the Gaussian distribution to reflect the sensitivity of sensor adaptability to environmental variables, where the adaptability score is at the best adaptability value. It reaches its maximum value near , and gradually decreases with the degree of deviation.

[0105] Weight Item Used to dynamically adjust the contribution of each environmental variable to the suitability score; parameter ∈ j is the fixed importance coefficient of the jth environmental variable, which is usually set by experience or prior knowledge. For example, in a scenario where humidity has a significant impact on the sensor, the humidity variable ∈ H will be assigned a higher value; R j It is the volatility or rate of change of the current environmental variables. For example, the amplitude of change is quantified by calculating the time derivative of humidity or terrain characteristics to reflect the dynamic impact of the variables on the current operating conditions.

[0106] The entire scoring mechanism uses a Gaussian distribution function to model the suitability of each environmental variable, combined with a dynamic weighting strategy to comprehensively consider the sensor's theoretical adaptability and the complexity of the real-time environment. This effectively avoids scoring bias caused by a single variable and allows flexible adjustments to adaptability assessment results when multiple variables change simultaneously, providing a scientific basis for further optimization of sensor weights. This approach can significantly improve the sensor's adaptability to changing environments when applied to complex agricultural scenarios, thereby supporting the high-precision and high-stability requirements of speed monitoring and control.

[0107] The above-mentioned step S202 uses the Kalman filter algorithm to fuse the three speed data to obtain the second predicted speed V2, including the following steps S401-S405:

[0108] Step S401: define the state variable of the system as the current speed prediction value x k , the measurement vector is the three speed data z provided by the sensor k =[a1,a2,a3] T , state transition matrix A k =1, control variable matrix B k = 0, the conversion matrix H from the state vector to the measurement vector k =[1,1,1] T / 3, system interference covariance matrix Q, and measurement interference covariance matrix R;

[0109] The state equation (prediction equation) of the system is: k =A k x k-1 +B k u k +w k ;

[0110] The measurement equation is: k =H k x k +v k ;

[0111] Where: u k is the state control vector; w k is the control system interference factor, which obeys the Gaussian distribution w k ~N(0,Q);v k is the measured interference factor, which obeys the Gaussian distribution v k ~N(0,R);

[0112] Step S402: Initialize the initial value of the state variable is the weighted average of the historical speed data or the first predicted speed V1, and the initialization state covariance matrix P0 is the estimated uncertainty;

[0113] Step S403: At each time step k, first predict the velocity according to the state transition equation and calculate the prior state estimate at the current moment:

[0114]

[0115] And calculate the prior state covariance:

[0116]

[0117] Step S404: Calculate the Kalman gain according to the measurement equation:

[0118]

[0119] Use the Kalman gain to update the posterior state estimate of the current velocity:

[0120]

[0121] And update the posterior state covariance:

[0122]

[0123] in: and The posterior state estimates at time k-1 and time k are respectively one of the filtering results, that is, the updated result, also called the best estimate; represents the prior state estimate at time k, and the state at time k is predicted based on the optimal estimate at time k-1; P k-1 and P k , respectively, the posterior estimated covariance at time k-1 and time k, which corresponds to and The covariance of represents the uncertainty of the state and is one of the results of filtering; is the prior estimated covariance at time k, which corresponds to Covariance of K k is the Kalman gain;

[0124] Step S405: Estimate the posterior state of each time step As the second predicted speed V2 after fusion.

[0125] A Kalman filter is used to fuse multi-sensor speed data. Through an iterative process of state prediction and observation update, the effects of sensor noise and system interference are effectively reduced, enhancing the smoothness and stability of speed prediction. This method fully leverages the complementary nature of encoder, GPS, and Doppler radar data to accurately estimate dynamic speed changes, improving the accuracy and real-time performance of speed monitoring and making it suitable for complex agricultural machinery operating environments.

[0126] The above-mentioned step S203 uses the DS evidence theory algorithm to fuse the first predicted speed V1 and the second predicted speed V2, and uses the fused speed prediction result V as the operating speed of the agricultural machinery, including the following steps S501-S505:

[0127] Step S501, define the recognition framework Θ = {V1, V2}, and construct the power set 2 Θ ={Φ,{V1},{V2},{V1,V2}}; the power set contains all possible subsets, including the empty set Φ, V1 alone, V2 alone, and the union of V1 and V2;

[0128] Step S502: define basic trust assignment functions m1 and m2 for each subset within the identification framework Θ, where m1({V1}) and m2({V2}) represent the direct trust values ​​for V1 and V2 respectively, and m1({V1, V2}) and m2({V1, V2}) represent the trust values ​​that hold for both, reflecting the uncertainty part. The basic trust allocation function satisfies the following conditions:

[0129]

[0130] in, That is to say and m(A) is the basic trust allocation function corresponding to subset A. A is a non-empty subset. A can be {V1}, {V2}, or {V1, V2}. Based on the above conditions, it is not clear how the trust values ​​of each non-empty subset are allocated. If the value of m(A) is greater than 0, then A can be called a focal element.

[0131] Step S503: Calculate the conflict coefficient K to measure the degree of conflict between V1 and V2, that is, V1 and V2 are two pieces of evidence in the DS evidence theory algorithm; the formula is:

[0132]

[0133] Among them: A i and B j Represent the focal elements of the basic trust allocation functions m1 and m2 respectively;

[0134] Step S504: Dempster's combination rule is used to calculate the value of the fused basic trust allocation function m(A) for each non-empty subset A:

[0135]

[0136] Step S505: Calculate the final speed prediction result V based on the fused basic trust allocation function m(A):

[0137] - If the confidence value of {V1} or {V2} is the largest, the corresponding predicted speed is selected as the speed prediction result V;

[0138] -If the trust value of {V1, V2} is the largest, V1 and V2 are further weighted and their weighted average is taken as the final speed prediction result V.

[0139] The above is only a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principles of the present invention. These improvements and modifications should also be regarded as the scope of protection of the present invention.

Claims

1. A method for monitoring the operating speed of intelligent agricultural machinery based on multi-sensor information fusion, the method comprising: Acquire three types of speed data: first speed data a1 measured by an encoder speed measuring device, second speed data a2 measured by a GPS speed measuring device, and third speed data a3 measured by a Doppler radar speed measuring device; The operating speed of agricultural machinery is obtained based on the three speed data; It is characterized in that: the agricultural machinery operating speed obtained based on the three speed data includes: The speed data are processed using a weighted average filtering algorithm, and the three speed data are fused to obtain the first predicted speed V1; The three speed data are fused using the Kalman filter algorithm to obtain the second predicted speed V2; Performing data fusion on the first predicted speed V1 and the second predicted speed V2 using the DS evidence theory algorithm, and using the fused speed prediction result V as the operating speed of the agricultural machinery; The weighted average filtering algorithm is used to process each speed data, and the three speed data are fused to obtain the first predicted speed V1, which includes: Get the speed data a corresponding to each i (t) time series, respectively denoted as A i ={a i (t1), a i (t2),…a i (t n )}; Where t1, t2, …, t n is the time point; i=1,2,3; Definition corresponds to the time series A i The weight of each element in ω(j); Use the following formula to get the processed speed data corresponding to the current moment: ； Where c is the current time point, j is the index of the historical data point in the weighted moving sliding window, and the sliding window only contains the current time point and the previous k data points; The current first predicted speed V1 is obtained by weighted calculation based on the processed three speed data: ； Among them, α, β, γ are weight coefficients; The first predicted speed is obtained by performing weighted calculation based on the processed three speed data. In the process of dynamically adjusting the weight coefficient, the specific steps include: Step A1: Initialize weights α0, β0, and γ0 for the encoder, GPS, and Doppler radar based on the theoretical accuracy of the sensor and typical applicable scenarios. The initial weights satisfy the normalization condition: α0+β0+γ0=1. Step A2: Collect the velocity data a1(t), a2(t), and a3(t) of each sensor, and use the sliding window to calculate the standard deviation σ of each sensor within the window. i : ； in: is the mean of the velocity data in the window, k is the window size, σ i The smaller the value, the higher the stability. Step A3, based on Bayesian inference, combines historical measurement performance with current operating conditions to calculate the credibility P i : ； Where: E i Condition; P(a i |E i ) is the probability density of the speed measured in the current environment, P(E i ) is the prior probability of the environmental condition, P(a i ) is the marginal probability of the sensor velocity data; Step A4, using external auxiliary sensor data) calculate the suitability score S of each sensor i : S i =f(H ,T,C,P); Function f is an empirical model or a training-based applicability calculation formula for specific environmental variables; H is humidity, T is terrain characteristics, C is temperature, and P is vibration intensity; Step A5: Integrate the stability, credibility and applicability scores to calculate the comprehensive score W for each sensor. i : ； Among them, λ1, λ2, and λ3 are proportional factors for adjusting the contribution of the three weights, which are adjusted according to the actual scenario; Step A6: Normalize the comprehensive score and update the weights α, β, and γ of each sensor: ； ； ； Step A7: Perform weighted fusion on the speed data using the updated weights to generate a first predicted speed V1 at the current moment.

2. The intelligent agricultural machinery operation speed monitoring method based on multi-sensor information fusion according to claim 1 is characterized in that: The method of fusing the three speed data using the Kalman filter algorithm to obtain the second predicted speed V2 includes: Define the state variable of the system as the current speed prediction value x k , the measurement vector is the three speed data z provided by the sensor k =[a1,a2,a3] T , state transition matrix A k =1, control variable matrix B k = 0, the conversion matrix H from the state vector to the measurement vector k =[1,1,1] T / 3, system interference covariance matrix Q, and measurement interference covariance matrix R; Initialize the initial value of the state variable , initialize the state covariance matrix P0; At each time step k, the velocity is first predicted according to the state transfer equation, and the prior state estimate at the current moment is calculated: ； And calculate the prior state covariance: ； The Kalman gain is calculated from the measurement equation: ； Use the Kalman gain to update the posterior state estimate of the current velocity: ； And update the posterior state covariance: ； in: and The posterior state estimates P at time k-1 and time k are k-1 and P k , respectively, the posterior estimated covariances at time k-1 and time k; is the prior estimated covariance at time k; K k is the Kalman gain; The posterior state estimate for each time step As the second predicted speed V2 after fusion.

3. The intelligent agricultural machinery operation speed monitoring method based on multi-sensor information fusion according to claim 1 is characterized in that: The method of fusing data of the first predicted speed V1 and the second predicted speed V2 by using the DS evidence theory algorithm and using the fused speed prediction result V as the operating speed of the agricultural machinery includes: Define the recognition framework Θ = {V1, V2} and construct the power set 2 Θ ={Φ,{V1},{V2},{V1,V2}}; Define basic trust assignment functions m1 and m2 for each subset within the identification framework Θ, where m1({V1}) and m2({V2}) represent the direct trust values ​​for V1 and V2 respectively, and m1({V1, V2}) and m2({V1, V2}) represent the trust values ​​that hold for both, reflecting the uncertainty part. = ; The basic trust allocation function satisfies the following conditions: ； in, That is to say and , m(A) is the basic trust allocation function corresponding to subset A, A is a non-empty subset; Calculate the conflict coefficient K to measure the degree of conflict between V1 and V2; the formula is: ； Among them: A i and B j Represent the focal elements of the basic trust allocation functions m1 and m2 respectively; Use Dempster's combination rule to calculate the value of the fused basic trust allocation function m(A) for each non-empty subset A: ； According to the fused basic trust distribution function m(A), the final speed prediction result V is calculated.

4. Intelligent agricultural machinery operation speed monitoring system based on multi-sensor information fusion, characterized by: It includes an encoder speed measuring device, a GPS speed measuring device and a Doppler radar speed measuring device, and also includes a controller connected to the three speed measuring devices. The controller can implement the intelligent agricultural machinery operation speed monitoring method based on multi-sensor information fusion as described in any one of claims 1-3.

Images