Digital signal transmission system and method thereof

Abstract

The invention relates to the technical field of data processing, in particular to a digital signal transmission system and method, and the method comprises the steps: obtaining agricultural data, and obtaining the periodic performance and monotonous trend performance of each monitoring project; clustering based on the periodical performance and the monotonous trend performance of all the monitoring items to obtain a plurality of data time sequence feature clusters; obtaining the segmentation length of the data time sequence feature cluster according to the deviation condition of the periodical representation and the monotonous trend representation of all monitoring items in the data time sequence feature cluster in combination with the distribution condition of the data to be compressed; constructing a plurality of coding dictionaries of each monitoring item in a segmentation manner according to the segmentation length; and matching each monitoring item in the next acquisition period to obtain an optimal matching coding dictionary, and compressing the optimal matching coding dictionary to obtain compressed data. The invention aims to solve the problem of large redundant data of the coding dictionary caused by the change of agricultural data along with the growth stage of the plant, and achieves the purposes of improving the compression efficiency and reducing the traversal complexity of the coding dictionary.

Description

Technical Field

[0001] The present invention relates to the technical field of data processing, and in particular to a digital signal transmission system and method thereof. Background Art

[0002] In the process of agricultural production and plant growth status monitoring, a large amount of information needs to be monitored and recorded. The sensors that monitor various monitoring items obtain a large amount of digital signals by converting the collected analog signals into digital signals. Summarizing and storing agricultural data is of great help in analyzing crop status and environment. The transmission of digital signals is a key link in modern communication systems, involving the process of transmitting information from the sender to the receiver. Since the bandwidth of the actual transmission medium of the agricultural management system is limited, the collected data needs to be compressed and transmitted to improve transmission efficiency and save storage space.

[0003] Using a compression algorithm based on a coding dictionary for agricultural data is one of the most common compression methods. For example, the LZ77 algorithm constructs a coding dictionary by constructing a traversal window to traverse the data, and uses a triple consisting of an offset value, a matching length, and the first character to be encoded in the encoding area in the coding dictionary to represent the compressed character; however, in the process of agricultural production management, as the growth stage of the plant changes, there are stage differences in the agricultural data obtained at different plant growth stages. When the agricultural data is compressed using a coding dictionary, the compression dictionary constructed based on the initial plant growth stage is no longer applicable to the data in the later plant growth stage. Therefore, more triplets need to be added to the dictionary, causing the coding dictionary to be bloated. In addition, the compression dictionary constructed based on the initial plant growth stage may not be used in the later plant growth stage, but it will still be traversed during encoding, resulting in reduced query efficiency of the triplets in the coding dictionary during compression and decompression. Summary of the invention

[0004] The present invention provides a digital signal transmission system and method thereof to solve the existing problems.

[0005] A digital signal transmission system and method of the present invention adopts the following technical solutions:

[0006] An embodiment of the present invention provides a digital signal transmission system and method thereof, the method comprising the following steps:

[0007] Acquire agricultural data, wherein the agricultural data includes all data to be compressed within a collection period of several monitoring projects;

[0008] According to the changing trend and periodicity of each monitoring item during the collection period, the periodicity performance and monotonic trend performance of each monitoring item are obtained;

[0009] Based on the periodic performance and monotonic trend performance of all monitoring items, clustering is performed to obtain several data time series feature clusters;

[0010] According to the periodic performance and monotonic trend performance bias of all monitoring items in the data time series feature cluster, the stage fault tolerance weight of the data time series feature cluster is obtained;

[0011] Taking the stage fault tolerance weight of the data time series feature cluster as the weight and combining the distribution of the data to be compressed of the same monitoring project in the data time series feature cluster, the segment length of the data time series feature cluster is obtained;

[0012] Segment each monitoring item in the data time series feature cluster according to the segment length, and construct several coding dictionaries for each monitoring item;

[0013] Match the distribution of the data to be compressed for each monitoring project in the next collection cycle with all the coding dictionaries of the same monitoring project to obtain the optimal matching coding dictionary for each monitoring project in the next collection cycle;

[0014] According to the distribution difference between the optimal matching coding dictionary of the monitoring items in the next collection cycle and the data to be compressed, the optimal matching coding dictionary is selected for compression to obtain all compressed data of all monitoring items in the agricultural data in each collection cycle, and the compressed data is transmitted for application in agricultural management.

[0015] Preferably, the specific steps for obtaining the periodic performance and monotonic trend performance of the monitoring items are:

[0016] For the ith monitoring item in the first collection cycle, all the data to be compressed of the ith monitoring item in the first collection cycle are input into the Mann-Kendall trend test model, and the standard normal distribution statistic of the ith monitoring item is recorded as the monotonic trend performance of the ith monitoring item;

[0017] Construct a frequency spectrum of all data to be compressed for the i-th monitoring item in the first acquisition cycle, and arrange the normalized amplitudes of all frequencies in the frequency spectrum of the i-th monitoring item in the first acquisition cycle in descending order after maximum and minimum normalization to obtain a descending sequence of amplitudes for the i-th monitoring item;

[0018] The periodic performance f of the i-th monitoring item in the first collection cycle i The calculation method is:

[0019]

[0020] Among them, n i is the number of frequencies in the period matching set of the i-th monitoring item, N iN' is the cumulative sum of the amplitudes of all frequencies of the i-th monitoring item, except the frequencies included in the period matching set of the i-th monitoring item. i is the cumulative sum of the amplitudes of all frequencies in the period matching set of the i-th monitoring item, and ε represents the preset hyperparameter.

[0021] Preferably, the specific method of obtaining the period matching set is:

[0022] The method for obtaining the period matching set of the i-th monitoring item in the first collection period is:

[0023] M i =FILTER{m i,k ,m i,k -m i,k+1 ≥T}

[0024] Among them, M i is the period matching set of the i-th monitoring item, m i,k is the normalized amplitude of the kth frequency in the descending sequence of the amplitude of the ith monitoring item, m i,k+1 is the normalized amplitude of the kth frequency in the amplitude descending sequence of the ith monitoring item, T is the preset threshold; and the FILTER{} function is a conditional selection function.

[0025] Preferably, the specific method of obtaining the stage fault tolerance weight is:

[0026] The monotonic trend performance of the i'th monitoring item in the jth data time series feature cluster is recorded as S' j,i' , the stage fault tolerance weight ω of the jth data time series feature cluster j The calculation method is:

[0027]

[0028] Among them, f' j,i' is the periodic performance of the i'th monitoring item in the jth data time series feature cluster, is the mean of the periodic performance of all monitoring items in the jth data time series feature cluster, max{} represents the maximum value function, and min{} represents the minimum value function.

[0029] Preferably, the step of acquiring the segment length of the data time series feature cluster includes:

[0030] The first data to be compressed in the i'th monitoring project of the j'th data time series feature cluster is recorded as the segmented starting data of the first stage of the i'th monitoring project;

[0031] For the ath data to be compressed in the i'th monitoring item of the jth data time series feature cluster, a stage division threshold is preset. When the stage end evaluation coefficient of the ath data to be compressed in the i'th monitoring item of the jth data time series feature cluster is less than the stage division threshold, a local range of the a+1th data to be compressed is constructed, and the stage end evaluation coefficient of the a+1th data to be compressed is calculated;

[0032] And so on, until there is a constant c, which makes the phase end evaluation coefficient of the a+cth data to be compressed greater than the phase division threshold, the data interval between the segmented start data of the first phase of the i'th monitoring project and the a+cth data to be compressed is recorded as the phase length of the first phase of the i'th monitoring project in the jth data time series feature cluster;

[0033] Among the stage lengths of all stages of all monitoring items of the jth data time series feature cluster, the largest stage length is recorded as the segment length of the jth data time series feature cluster.

[0034] Preferably, the specific method of obtaining the evaluation coefficient at the end of the stage is:

[0035] The phase end evaluation coefficient G of the a-th data to be compressed in the i'th monitoring item of the j-th data time series feature cluster j,i',a The calculation method is:

[0036] G j,i',a =sigmoid{(1-ω j )*|M j,i' -M' j,i' |}

[0037] Among them, ω j is the stage fault tolerance weight of the jth data time series feature cluster, M j,i' is the mean of all the data to be compressed in the local range of the ath data to be compressed in the i'th monitoring item of the jth data time series feature cluster, M' j,i' is the mean value of the segmented starting data of the first stage of the i'th monitoring item to the previous data to be compressed of the a'th data to be compressed in the i'th monitoring item in the j'th data time series feature cluster, || represents the absolute value function, and sigmoid{} represents the Sigmoid function.

[0038] Preferably, the specific method of obtaining the local range is:

[0039] The local range of the ath data to be compressed is obtained by selecting the local range of the ath data to be compressed in the i'th monitoring item of the jth data time series feature cluster from the starting point of the local range and moving toward the ath data to be compressed in time sequence with a preset local range length.

[0040] Preferably, the optimal matching coding dictionary of the monitoring items includes:

[0041] For the kth monitoring item, the mean of all data in each coding dictionary of the kth monitoring item is used as the anchor point, the kurtosis of all data is used as the cluster distribution coefficient, and the variance of all data is used as the secondary distribution coefficient;

[0042] The anchor points, cluster distribution coefficients and secondary distribution coefficients of all coding dictionaries of the kth monitoring project are used as the x-axis, y-axis and z-axis respectively to construct a matching space to obtain the matching space points of each coding dictionary;

[0043] For the data to be compressed in the next acquisition cycle of the kth monitoring project, the data are divided according to the stage length of the data time series feature cluster to obtain several data segments to be compressed in the next acquisition cycle of the kth monitoring project. The anchor point, cluster distribution coefficient and secondary distribution coefficient of each data segment to be compressed in the next acquisition cycle of the kth monitoring project are calculated respectively to obtain the spatial point to be matched. After mapping to the matching space, the coding dictionary corresponding to the matching space point closest to the spatial point to be matched is selected as the optimal matching coding dictionary for each data segment to be compressed in the next acquisition cycle of the kth monitoring project.

[0044] Preferably, the specific method of obtaining the compressed data is:

[0045] A deviation threshold is preset, and for the p-th data segment to be compressed in the next acquisition cycle of the k-th monitoring item, when the normalized value of the Euclidean distance between the to-be-matched spatial point of the p-th data segment to be compressed and the matching spatial point of the optimal matching coding dictionary in the matching space is less than or equal to the preset deviation threshold, the p-th data segment to be compressed is encoded and compressed using the optimal matching coding dictionary of the p-th data segment to be compressed to obtain compressed data;

[0046] When the normalized value of the Euclidean distance between the to-be-matched spatial point of the p-th data segment to be compressed and the matching spatial point of the optimal matching coding dictionary in the matching space is greater than a preset deviation threshold, a new coding dictionary is constructed for the p-th data segment to be compressed to obtain the compressed data of the p-th data segment to be compressed.

[0047] The present invention also proposes a digital signal transmission system, which includes a memory, a processor, and a computer program stored in the memory and running on the processor, and the processor implements the steps of the above method when executing the computer program.

[0048] The beneficial effects of the technical solution of the present invention are as follows: the present invention obtains the periodic performance and monotonic trend performance of each monitoring item according to the change trend and periodicity of each monitoring item of agricultural data within the collection period, and then performs clustering to obtain a number of data time series feature clusters, so as to achieve the goal of clustering related monitoring items into one category according to the periodicity and change amplitude of different monitoring items, so that the data of monitoring items in the same stage are associated, and the data of monitoring items in one stage are avoided to be divided into multiple different stages due to errors and the like; according to the bias of the periodic performance and monotonic trend performance of all monitoring items in the data time series feature cluster, combined with the distribution of the data to be compressed of the same monitoring item in the data time series feature cluster, the segment length of the data time series feature cluster is obtained, so as to avoid the influence of the fluctuation of the periodic data on the division of the stage of the data to be compressed, and also achieve the purpose of uniformly dividing the stages of the monitoring items in the same data time series feature cluster, thereby reducing the number of encoding dictionaries and encoding complexity; according to the segment length, the data time series feature cluster is divided into different stages according to the segment length. Each monitoring item in the feature cluster is segmented, and several coding dictionaries are constructed for each monitoring item. The distribution of the data to be compressed for each monitoring item in the next acquisition cycle is matched with all coding dictionaries of the same monitoring item, and the optimal matching coding dictionary for each monitoring item in the next acquisition cycle is obtained, so as to match the optimal coding dictionary based on the collected data and the existing coding dictionary, and use the existing coding dictionary as the prior coding dictionary to reduce the complexity of dictionary construction; according to the distribution difference between the optimal matching coding dictionary and the data to be compressed for the monitoring item in the next acquisition cycle, the optimal matching coding dictionary is selected for compression, and all compressed data of all monitoring items in the agricultural data in each acquisition cycle are obtained, and the difference between the matching coding dictionary and the actual collected data is judged to select whether to use the optimal matching coding dictionary, thereby reducing the inability of the inappropriate coding dictionary to provide reference for encoding, and finally achieving the purpose of improving compression efficiency when the coding dictionary is used for data compression and transmission. BRIEF DESCRIPTION OF THE DRAWINGS

[0049] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0050] Figure 1 The present invention is a flowchart of the steps of a digital signal transmission method. DETAILED DESCRIPTION

[0051] In order to further explain the technical means and effects adopted by the present invention to achieve the predetermined invention purpose, the following is a detailed description of a digital signal transmission system and method thereof proposed by the present invention, its specific implementation, structure, features and effects, in conjunction with the accompanying drawings and preferred embodiments. In the following description, different "one embodiment" or "another embodiment" does not necessarily refer to the same embodiment. In addition, specific features, structures or characteristics in one or more embodiments may be combined in any suitable form.

[0052] Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs.

[0053] The following describes in detail a digital signal transmission system and method provided by the present invention in conjunction with the accompanying drawings.

[0054] See also Figure 1 , which shows a flow chart of a method for transmitting a digital signal provided by an embodiment of the present invention, the method comprising the following steps:

[0055] Step S001, obtaining agricultural data.

[0056] The purpose of this embodiment is to analyze the agricultural data, thereby constructing a coding dictionary for the collected agricultural data according to the differences in plant growth stages, and then matching the coding dictionary for the agricultural data collected at other plant growth stages or constructing a new coding dictionary. Therefore, it is necessary to first collect the agricultural data to be encoded and compressed.

[0057] It should be noted that a certain amount of agricultural data is required as a data basis when constructing the initial coding dictionary, and the analysis of agricultural data and data feature extraction require data collected over a long period of time as support, while the real-time requirements for data transmission are not high. Therefore, this embodiment realizes agricultural data compression by analyzing the agricultural data obtained in each collection cycle and constructing a coding dictionary.

[0058] Preferably, in this embodiment, during each collection cycle, agricultural data including soil data, air environment data, plant data, agricultural management data, etc. are obtained at different sampling intervals by combining different monitoring items with corresponding sensors or chemical inspection methods. The monitoring items of soil data mainly include soil moisture, soil temperature, pH value, and nutrient content; the monitoring items of air environment data mainly include air temperature, air humidity, precipitation, wind direction and speed, and air pressure; the monitoring items of plant data mainly include plant height and plant maturity; and the monitoring items of agricultural management data mainly include fertilizer usage and agricultural machinery usage.

[0059] Furthermore, after obtaining all agricultural data of all monitoring items within a collection cycle, linear interpolation is performed on the agricultural data of all monitoring items with the minimum sampling interval in all monitoring items to supplement the incomplete values, and all supplemented agricultural data are normalized to the maximum and minimum values ​​to obtain the data to be compressed in each collection cycle.

[0060] Step S002: Obtain the periodicity and monotonic trend performance of each monitoring item according to the change trend and periodicity of each monitoring item within the collection period; perform clustering based on the periodicity and monotonic trend performance of all monitoring items to obtain several data time series feature clusters.

[0061] It should be noted that plants need to be applied with fertilizers with different nutrients and different amounts of water at different growth stages, and fertilizers need to be replenished in time when fertility declines. Therefore, the data collected for some monitoring items undergo periodic changes with changes in soil composition and nutrient composition, and the plant data shows a monotonic trend over time, resulting in the collected monitoring items including trend monitoring items and periodic monitoring items. When compressing the data of monitoring items with different data characteristics, the periodic monitoring items contain more repetitive data, and the amount of data that can be compressed by a coding dictionary is greater than that of trend monitoring items. Therefore, it is necessary to first distinguish between periodic monitoring items and monotonic trend monitoring items.

[0062] Preferably, for all monitoring items in the first collection cycle, the time series characteristics of the data to be compressed for all monitoring items in the first collection cycle are divided according to the monotonic trend expression and periodic expression of the data to be compressed in each monitoring item, and a plurality of data time series characteristic clusters are obtained. The specific steps are:

[0063] The Mann-Kendall trend test model obtains the standard normal distribution statistic of the analyzed data by analyzing the degree of change of the analyzed data in a specific order. The value interval of the standard normal distribution statistic is [-1,1]. When the standard normal distribution statistic approaches 1, it means that the analyzed data presents a monotonically increasing trend. When the standard normal distribution statistic approaches -1, it means that the analyzed data presents a monotonically decreasing trend. When the standard normal distribution statistic approaches 0, it means that the analyzed data does not have an obvious monotonic trend.

[0064] Furthermore, for the i-th monitoring item in the first collection cycle, all the data to be compressed of the i-th monitoring item in the first collection cycle are input into the Mann-Kendall trend test model, and the standard normal distribution statistic of the i-th monitoring item is recorded as the monotonic trend expression of the i-th monitoring item.

[0065] Furthermore, since the amount of nutrients required by plants at different growth stages is different, as the plants grow, their absorption efficiency becomes greater than that of the seedling stage, which will accelerate the loss of nutrients in the soil, thereby shortening the fertilization cycle. The change cycle of each nutrient in the collected monitoring items and the cycle of the fertilization cycle data change, and the cycle of the same monitoring item will change with the different growth stages of the plants. If a single cycle is used to measure the periodic changes of the monitoring items, there is a situation where there is a significant difference between the selected cycle and the actual cycle of the monitoring item, resulting in errors in the cycle analysis; therefore, it is first necessary to obtain the cycle corresponding to the different data to be compressed for each monitoring item.

[0066] Preferably, for the ith monitoring item in the first acquisition cycle, a frequency spectrum of all data to be compressed for the ith monitoring item in the first acquisition cycle is constructed, the frequency spectrum containing several frequencies and the amplitude of each frequency, if there is a frequency in the frequency spectrum whose amplitude is much larger than the amplitude of other frequencies, then it means that the frequency is a cycle corresponding to the ith monitoring item, and for multi-stage data to be compressed, it may correspond to multiple frequencies with amplitudes much larger than other frequencies in the frequency spectrum, then these frequencies are the corresponding cycles of the ith monitoring item. Therefore, this embodiment constructs a cycle matching set for the ith monitoring item in the first acquisition cycle by judging the frequencies in the frequency spectrum of the ith monitoring item that satisfy the condition that the amplitude is much larger than other frequencies.

[0067] Furthermore, according to the normalized amplitudes of all frequencies in the spectrum diagram of the i-th monitoring item in the first acquisition cycle after maximum and minimum normalization, the amplitude descending sequence of the i-th monitoring item is obtained by arranging them in descending order. Then, in the first acquisition cycle, the period matching set of the i-th monitoring item is obtained as follows:

[0068] M i =FILTER{m i,k ,m i,k -m i,k+1 ≥T}

[0069] Among them, M i is the period matching set of the i-th monitoring item, m i,k is the normalized amplitude of the kth frequency in the descending sequence of the amplitude of the i-th monitoring item, m i,k+1 is the normalized amplitude of the kth frequency in the amplitude descending sequence of the i-th monitoring item, T is the preset threshold, and this embodiment is described by taking 0.2 as an example; FILTER{} function is a conditional selection function;

[0070] FILTER{m i,k ,m i,k -m i,k+1<T> represents that when the normalized amplitudes of the k-th frequency and the (k - 1)-th frequency in the descending amplitude sequence of the i-th monitoring item satisfy m i,k -m i,k+1 <T, the normalized amplitude m of the k-th frequency in the descending amplitude sequence of the i-th monitoring item i,k is added to the period matching set of the i-th monitoring item, otherwise the function stops traversing the descending amplitude sequence of the i-th monitoring item.

[0071] Furthermore, by analyzing the difference between the amplitude in the period matching set of the i-th monitoring item and other amplitudes, the period performance of the i-th monitoring item in the first acquisition period is obtained.

[0072] The period performance f of the i-th monitoring item in the first acquisition period i is calculated as follows:

[0073]

[0074] where n i is the number of frequencies in the period matching set of the i-th monitoring item, N i is the sum of the amplitudes of all frequencies of the i-th monitoring item except those included in the period matching set of the i-th monitoring item, N' i is the sum of the amplitudes of all frequencies in the period matching set of the i-th monitoring item, and ε represents a preset hyperparameter, which is described by taking 0.1 as an example in this embodiment and is used to avoid the denominator being 0.

[0075] The fewer the number of frequencies in the period matching set of the i-th monitoring item, the fewer the number of frequencies that satisfy the amplitude being much larger than other frequencies, and the more prominent the significant peak characteristics of the period corresponding to the frequencies in the period matching set in the spectrogram. The greater the difference between the sum of the amplitudes of all frequencies in the period matching set of the i-th monitoring item and the amplitudes of the remaining frequencies, the more prominent the significant peak characteristics of the period corresponding to the frequencies in the period matching set in the spectrogram.

[0076] Similarly, the period performance and the monotonic trend performance of each monitoring item in the first acquisition period are obtained, and a two-dimensional clustering space is constructed, where the horizontal axis is the period performance and the vertical axis is the monotonic trend performance. The period performance and the monotonic trend performance of each monitoring item in the first acquisition period are mapped into the two-dimensional clustering space to obtain those of each monitoring item. After setting the preset initial number of clusters, the ISODATA clustering algorithm is used to cluster the two-dimensional clustering space to obtain several data time series feature clusters. The initial number of clusters of the ISODATA clustering algorithm is described by taking 3 as an example, and other values can be set in other embodiments, which are not specifically limited in this embodiment.

[0077] Step S003: According to the periodic performance and monotonic trend performance bias of all monitoring items in the data time series feature cluster, combined with the distribution of the data to be compressed for the same monitoring item in the data time series feature cluster, obtain the segment length of the data time series feature cluster; segment each monitoring item in the data time series feature cluster according to the segment length, and construct several coding dictionaries for each monitoring item.

[0078] It should be noted that there is a certain correlation between different monitoring items in agricultural data. For example, soil temperature and humidity will change with changes in air temperature, and the content of soil phosphorus and potassium elements will change with the rate of change of plant growth height, resulting in changes in the size of the cycle. Therefore, the changes in the cycle and trend of monitoring items in the same data time series feature cluster are approximately uniform. In this way, the monitoring items in the same data time series feature cluster are further analyzed in stages to obtain the stage length of each monitoring item in each cluster, so that the data of monitoring items in the same stage are correlated, avoiding the data of monitoring items in one stage from being divided into multiple different stages due to errors and other reasons, resulting in too many coding dictionaries being constructed, the number of triplets in each coding dictionary being too small, resulting in reduced data compression efficiency, and too much space occupied by the coding dictionary.

[0079] It should be further explained that the monitoring items in the agricultural data may have both monotonic trend changes and periodic changes. Therefore, when the growth stages are divided by the similarity of the data distribution of the data values ​​of the latter segment of the data to be compressed and the previous segment of the data to be compressed, the periodic changes of the data to be compressed cause errors in the range of the data values ​​to be compressed, resulting in the periodic changes of the data to be compressed belonging to the same stage being divided into multiple stages during the stage division. Therefore, this embodiment first obtains the stage fault tolerance weights of all monitoring items in each data time series feature cluster based on the bias of the periodic expression and monotonic trend expression of all monitoring items in each data time series feature cluster.

[0080] Preferably, the stage fault tolerance weight ω of the jth data time series feature cluster is j The calculation method is:

[0081]

[0082] Among them, S' j,i' is the monotonic trend performance of the i'th monitoring item in the jth data time series feature cluster, f' j,i' is the periodic performance of the i'th monitoring item in the jth data time series feature cluster, is the mean of the periodic performance of all monitoring items in the jth data time series feature cluster, max{} represents the maximum value function, and min{} represents the minimum value function.

[0083] When the jth data time series feature cluster satisfies max{S' j,i'}>min{f' j,i'}, indicating that the data of all monitoring items in the jth data time series feature cluster presents a monotonic trend feature, and the error in the range of the data values ​​to be compressed caused by periodicity is less affected. Therefore, the stage fault tolerance weight is reset to 0 when performing stage division, that is, the stage division is performed based on the data difference mainly caused by the trend change. On the contrary, when the jth data time series feature cluster satisfies max{S' j,i'}≤min{f' j,i'}, it means that the data of all monitoring items in the jth data time series feature cluster show periodic characteristics. Therefore, it is necessary to set the stage fault tolerance weight according to the degree of periodic performance of all monitoring items. The stronger the periodic performance, the more susceptible it is to the impact of data periodic changes when dividing the stages.

[0084] Furthermore, after obtaining the stage fault tolerance weight of each data time series feature cluster, the stage fault tolerance weight is used as the weight. When the monitoring project has obvious differences in the local range compared with all the previous data to be compressed, it is considered that the monitoring project has entered the next stage at this moment and needs to be divided into stages. Therefore, this embodiment traverses all the data to be compressed of the monitoring project, judges the difference between the data to be compressed in the local range and the previous data to be compressed, and obtains the stage end coefficient of each monitoring project in each data to be compressed in combination with the stage fault tolerance weight, and realizes the stage division of each monitoring project according to the stage end coefficient.

[0085] Preferably, as an example, this embodiment takes the i'th monitoring item of the j'th data timing feature cluster as an example for analysis, and records the first data to be compressed in the i'th monitoring item of the j'th data timing feature cluster as the segmented starting data of the first stage of the i'th monitoring item. For the a'th data to be compressed in the i'th monitoring item of the j'th data timing feature cluster, the a'th data to be compressed is taken as the starting point of the local range, and the local range of the a'th data to be compressed is selected according to the preset local range length in time sequence after the a'th data to be compressed. The local range length of this embodiment is described taking 100 data to be compressed as an example, and other embodiments may use other values, which are not specifically limited in this embodiment.

[0086] Then the phase end evaluation coefficient G of the a-th data to be compressed in the i'th monitoring item of the j-th data time series feature cluster is j,i',a The calculation method is:

[0087] G j,i',a =sigmoid{(1-ω j )*|M j,i' -M' j,i' |}

[0088] Among them, ω j is the stage fault tolerance weight of the jth data time series feature cluster, M j,i' is the mean of all the data to be compressed in the local range of the ath data to be compressed in the i'th monitoring item of the jth data time series feature cluster, M' j,i' is the mean value of the segmented starting data of the first stage of the i'th monitoring project to the previous data to be compressed of the a'th data to be compressed in the j'th monitoring project in the j'th data time series feature cluster, || represents the absolute value function, and sigmoid{} represents the Sigmoid function.

[0089] It should be noted that |M j,i' -M' j,i' | represents the stage difference between all the data to be compressed in the local range of the ath data to be compressed in the i'th monitoring project of the jth data time series feature cluster and the data to be compressed in the undivided stage. The larger the value, the greater the difference between the data to be compressed in the local range and the data to be compressed in the undivided stage.

[0090] Further, a stage division threshold is preset. In this embodiment, 0.68 is used as the stage division threshold. When the stage end evaluation coefficient of the a-th data to be compressed in the i'-th monitoring item of the j-th data time series feature cluster is less than the stage division threshold, it means that the difference between the data to be compressed in the local range and the data to be compressed in the undivided stage is not enough to divide the stage, and the local range of the a+1-th data to be compressed is constructed, and the stage end evaluation coefficient of the a+1-th data to be compressed is calculated;

[0091] And so on, until there is a constant c, which makes the stage end evaluation coefficient of the a+cth data to be compressed greater than the stage division threshold, indicating that the data after the a+cth data to be compressed is in a different stage from the previous data, and the data interval between the segmented start data of the first stage of the i'th monitoring project to the a+cth data to be compressed is recorded as the stage length of the first stage of the i'th monitoring project in the jth data time series feature cluster, and the a+c+1th data to be compressed is recorded as the segmented start data of the next stage;

[0092] Similarly, the stage lengths of all stages of each monitoring item of the jth data time series feature cluster are obtained. Since similar data changes are presented in a data time series feature cluster, the largest stage length among the stage lengths of all stages of all monitoring items of the jth data time series feature cluster is recorded as the segment length of the jth data time series feature cluster.

[0093] Furthermore, for each monitoring item of the jth data time series feature cluster, all data in the first acquisition cycle are phased-segmented using the segment length of the jth data time series feature cluster to obtain a number of segmented data to be compressed for each monitoring item in the jth data time series feature cluster, and a coding dictionary is constructed using the segmented data to be compressed to obtain a number of coding dictionaries for each monitoring item in the jth data time series feature cluster; the compression algorithm used in constructing the coding dictionary in this embodiment is the LZ77 algorithm, and other embodiments may be implemented using other compression algorithms based on coding dictionaries, which are not specifically limited in this embodiment, and the LZ77 algorithm to construct the coding dictionary is an existing well-known technology, which will not be repeated in this embodiment.

[0094] Similarly, in the first acquisition cycle, each data time series feature cluster is analyzed to obtain several coding dictionaries for each monitoring item.

[0095] Step S004: match the distribution of the data to be compressed for each monitoring project in the next collection cycle with all coding dictionaries of the same monitoring project to obtain the optimal matching coding dictionary for each monitoring project in the next collection cycle; based on the distribution difference between the optimal matching coding dictionary of the monitoring project in the next collection cycle and the data to be compressed, the optimal matching coding dictionary is selected for compression to obtain all compressed data of all monitoring projects in the agricultural data in each collection cycle.

[0096] It should be noted that since many coding dictionaries are constructed for each monitoring project, when the data to be compressed of the same monitoring project is obtained in other collection cycles, the coding dictionary compression algorithm mainly obtains the optimal matching coding dictionary based on the repetitiveness of the triples in the dictionary when compression is required. The optimal matching coding dictionary is used to compress the data to be compressed in other collection cycles, and the coding dictionary is supplemented according to the compression of the data to be compressed. Therefore, it is necessary to first obtain the optimal matching coding dictionary compressed in other collection cycles.

[0097] It should be further explained that when the distribution of triplet data in the optimal matching coding dictionary after matching is still very different from the distribution of the data to be compressed, using the optimal matching coding dictionary for encoding will result in more new triplets being added to the optimal matching coding dictionary, that is, the utilization rate of the triplets in the optimal matching coding dictionary is reduced, thereby reducing the compression efficiency. Therefore, it is necessary to determine whether to use the optimal matching coding dictionary for encoding compression or to construct a new coding dictionary for the compressed data based on the similarity between the data to be compressed and the data in the optimal matching coding dictionary.

[0098] Preferably, for the kth monitoring item, the mean of all data in each coding dictionary of the kth monitoring item is used as the anchor point to represent the data distribution situation of the coding dictionary, the kurtosis of all data is used as the clustering distribution coefficient to represent the data distribution clustering situation of the coding dictionary, and the variance of all data is used as the secondary distribution coefficient to represent the discrete situation of the data distribution of the coding dictionary.

[0099] Furthermore, the anchor points, clustered distribution coefficients and secondary distribution coefficients of all coding dictionaries of the k-th monitoring project are used as the x-axis, y-axis and z-axis respectively to construct a matching space to obtain the matching space points of each coding dictionary. For the data to be compressed in the next acquisition cycle of the k-th monitoring project, the data is divided according to the stage length of the data time series feature cluster, and several data segments to be compressed in the next acquisition cycle of the k-th monitoring project are obtained. The anchor points, clustered distribution coefficients and secondary distribution coefficients of each data segment to be compressed in the next acquisition cycle of the k-th monitoring project are calculated to obtain the space points to be matched. After mapping to the matching space, the coding dictionary corresponding to the matching space point closest to the space point to be matched is selected as the optimal matching coding dictionary for each data segment to be compressed in the next acquisition cycle of the k-th monitoring project.

[0100] Furthermore, a deviation threshold is preset. In this embodiment, 0.05 is used as the deviation threshold. For the p-th data segment to be compressed of the k-th monitoring item in the next acquisition cycle, when the normalized value of the Euclidean distance between the space point to be matched of the p-th data segment to be compressed and the matching space point of the optimal matching coding dictionary in the matching space is less than or equal to the preset deviation threshold, the optimal matching coding dictionary of the p-th data segment to be compressed is used to encode and compress the p-th data segment to be compressed to obtain compressed data; when the normalized value of the Euclidean distance between the space point to be matched of the p-th data segment to be compressed and the matching space point of the optimal matching coding dictionary in the matching space is greater than the preset deviation threshold, a new coding dictionary is constructed for the p-th data segment to be compressed to obtain the compressed data of the p-th data segment to be compressed, and the coding dictionary of the p-th data segment to be compressed is retained in the matching space to obtain a new matching space.

[0101] Similarly, all compressed data of each monitoring item in each collection cycle are obtained, and the matching space and the coding dictionary corresponding to each matching space point in the matching space are updated.

[0102] Furthermore, the compressed data of each data segment to be compressed and the corresponding coding dictionary are transmitted to the big data analysis model of agricultural management, and applied to agricultural management according to the analysis results of the big data analysis model.

[0103] Another embodiment of the present invention provides a digital signal transmission system, which includes a memory, a processor, and a computer program stored in the memory and running on the processor. When the processor executes the computer program, the above method steps S001 to S004 are implemented.

[0104] This is the end of the present invention.

[0105] 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 principles of the present invention should be included in the protection scope of the present invention.

Claims

1. A method for transmitting a digital signal, characterized in that: The method comprises the following steps: Acquire agricultural data, wherein the agricultural data includes all data to be compressed within a collection period of several monitoring projects; According to the changing trend and periodicity of each monitoring item during the collection period, the periodicity performance and monotonic trend performance of each monitoring item are obtained; Based on the periodic performance and monotonic trend performance of all monitoring items, clustering is performed to obtain several data time series feature clusters; According to the periodic performance and monotonic trend performance bias of all monitoring items in the data time series feature cluster, the stage fault tolerance weight of the data time series feature cluster is obtained; Taking the stage fault tolerance weight of the data time series feature cluster as the weight and combining the distribution of the data to be compressed of the same monitoring project in the data time series feature cluster, the segment length of the data time series feature cluster is obtained; Segment each monitoring item in the data time series feature cluster according to the segment length, and construct several coding dictionaries for each monitoring item; Match the distribution of the data to be compressed for each monitoring project in the next collection cycle with all the coding dictionaries of the same monitoring project to obtain the optimal matching coding dictionary for each monitoring project in the next collection cycle; According to the distribution difference between the optimal matching coding dictionary of the monitoring items in the next collection cycle and the data to be compressed, the optimal matching coding dictionary is selected for compression to obtain all compressed data of all monitoring items in the agricultural data in each collection cycle, and the compressed data is transmitted for application in agricultural management.

2. A method for transmitting a digital signal according to claim 1, characterized in that: The specific steps for obtaining the periodic performance and monotonic trend performance of the monitoring items are as follows: For the ith monitoring item in the first collection cycle, all the data to be compressed of the ith monitoring item in the first collection cycle are input into the Mann-Kendall trend test model, and the standard normal distribution statistic of the ith monitoring item is recorded as the monotonic trend performance of the ith monitoring item; Construct a frequency spectrum of all data to be compressed for the i-th monitoring item in the first acquisition cycle, and arrange the normalized amplitudes of all frequencies in the frequency spectrum of the i-th monitoring item in the first acquisition cycle in descending order after maximum and minimum normalization to obtain a descending sequence of amplitudes for the i-th monitoring item; The periodic performance f of the i-th monitoring item in the first collection cycle i The calculation method is: Among them, n i is the number of frequencies in the period matching set of the i-th monitoring item, N i N' is the cumulative sum of the amplitudes of all frequencies of the i-th monitoring item, except the frequencies included in the period matching set of the i-th monitoring item. i is the cumulative sum of the amplitudes of all frequencies in the period matching set of the i-th monitoring item, and ε represents the preset hyperparameter.

3. A method for transmitting a digital signal according to claim 2, characterized in that: The specific method of obtaining the period matching set is: The method for obtaining the period matching set of the i-th monitoring item in the first collection period is: M i =FILTER{m i,k ,m i,k -m i,k+1 ≥T} Among them, M i is the period matching set of the i-th monitoring item, m i,k is the normalized amplitude of the kth frequency in the descending sequence of the amplitude of the i-th monitoring item, m i,k+1 is the normalized amplitude of the kth frequency in the amplitude descending sequence of the ith monitoring item, T is the preset threshold; and the FILTER{} function is a conditional selection function.

4. A method for transmitting a digital signal according to claim 1, characterized in that: The specific method of obtaining the fault tolerance weight in the stage is: The monotonic trend performance of the i'th monitoring item in the jth data time series feature cluster is recorded as S' j,i' , the stage fault tolerance weight ω of the jth data time series feature cluster j The calculation method is: Among them, f' j,i' is the periodic performance of the i'th monitoring item in the jth data time series feature cluster, is the mean of the periodic performance of all monitoring items in the jth data time series feature cluster, max{} represents the maximum value function, and min{} represents the minimum value function.

5. The method for transmitting a digital signal according to claim 1, characterized in that: The step of obtaining the segment length of the data time series feature cluster includes: The first data to be compressed in the i'th monitoring project of the j'th data time series feature cluster is recorded as the segmented starting data of the first stage of the i'th monitoring project; For the ath data to be compressed in the i'th monitoring item of the jth data time series feature cluster, a stage division threshold is preset. When the stage end evaluation coefficient of the ath data to be compressed in the i'th monitoring item of the jth data time series feature cluster is less than the stage division threshold, a local range of the a+1th data to be compressed is constructed, and the stage end evaluation coefficient of the a+1th data to be compressed is calculated; And so on, until there is a constant c, which makes the phase end evaluation coefficient of the a+cth data to be compressed greater than the phase division threshold, the data interval between the segmented start data of the first phase of the i'th monitoring project and the a+cth data to be compressed is recorded as the phase length of the first phase of the i'th monitoring project in the jth data time series feature cluster; Among the stage lengths of all stages of all monitoring items of the jth data time series feature cluster, the largest stage length is recorded as the segment length of the jth data time series feature cluster.

6. A digital signal transmission method according to claim 5, characterized in that: The specific method of obtaining the evaluation coefficient at the end of the stage is: The phase end evaluation coefficient G of the a-th data to be compressed in the i'th monitoring item of the j-th data time series feature cluster j,i',a The calculation method is: G j,i',a =sigmoid{(1-ω j )*|M j,i' -M j,i' |} Among them, ω j is the stage fault tolerance weight of the jth data time series feature cluster, M j,i' is the mean of all the data to be compressed in the local range of the ath data to be compressed in the i'th monitoring item of the jth data time series feature cluster, M' j,i' is the mean value of the segmented starting data of the first stage of the i'th monitoring item to the previous data to be compressed of the a'th data to be compressed in the i'th monitoring item in the j'th data time series feature cluster, || represents the absolute value function, and sigmoid{} represents the Sigmoid function.

7. A digital signal transmission method according to claim 5, characterized in that: The specific method of obtaining the local range is: The local range of the ath data to be compressed is obtained by selecting the local range of the ath data to be compressed in the i'th monitoring item of the jth data time series feature cluster from the starting point of the local range and moving toward the ath data to be compressed in time sequence with a preset local range length.

8. The method for transmitting a digital signal according to claim 1, characterized in that: The optimal matching coding dictionary of the monitoring items includes: For the kth monitoring item, the mean of all data in each coding dictionary of the kth monitoring item is used as the anchor point, the kurtosis of all data is used as the cluster distribution coefficient, and the variance of all data is used as the secondary distribution coefficient; The anchor points, cluster distribution coefficients and secondary distribution coefficients of all coding dictionaries of the kth monitoring project are used as the x-axis, y-axis and z-axis respectively to construct a matching space to obtain the matching space points of each coding dictionary; For the data to be compressed in the next acquisition cycle of the kth monitoring project, the data are divided according to the stage length of the data time series feature cluster to obtain several data segments to be compressed in the next acquisition cycle of the kth monitoring project. The anchor point, cluster distribution coefficient and secondary distribution coefficient of each data segment to be compressed in the next acquisition cycle of the kth monitoring project are calculated respectively to obtain the spatial point to be matched. After mapping to the matching space, the coding dictionary corresponding to the matching space point closest to the spatial point to be matched is selected as the optimal matching coding dictionary for each data segment to be compressed in the next acquisition cycle of the kth monitoring project.

9. The method for transmitting a digital signal according to claim 1, characterized in that: The specific method of obtaining the compressed data is: A deviation threshold is preset, and for the p-th data segment to be compressed in the next acquisition cycle of the k-th monitoring item, when the normalized value of the Euclidean distance between the to-be-matched spatial point of the p-th data segment to be compressed and the matching spatial point of the optimal matching coding dictionary in the matching space is less than or equal to the preset deviation threshold, the p-th data segment to be compressed is encoded and compressed using the optimal matching coding dictionary of the p-th data segment to be compressed to obtain compressed data of the p-th data segment to be compressed; When the normalized value of the Euclidean distance between the to-be-matched spatial point of the p-th data segment to be compressed and the matching spatial point of the optimal matching coding dictionary in the matching space is greater than a preset deviation threshold, a new coding dictionary is constructed for the p-th data segment to be compressed to obtain compressed data.

10. A digital signal transmission system, comprising a memory, a processor, and a computer program stored in the memory and running on the processor, characterized in that: When the processor executes the computer program, the steps of the digital signal transmission method as described in any one of claims 1 to 9 are implemented.

Images